//! Test reports from OxygenFlow jobs (oxygenhub#114): a job's artifact

//! named `junit`, in JUnit XML as pytest `--junitxml`, cargo-nextest, Jest,

//! gotestsum and most runners write it, is parsed when it is uploaded and

//! kept per run: the counts, and every test case with its status and time.

//! A failure's message and output are kept, cut to a size a page can show.

use crate::db::{now_ms, Pool};

use crate::Shared;

use quick_xml::events::{BytesStart, Event};

use quick_xml::{Reader, XmlVersion};

/// The artifact name a job gives its JUnit report.

pub const ARTIFACT: &str = "junit";

/// Largest report read; a bigger one is noted in the job's log and skipped.

pub const MAX_REPORT: usize = 64 << 20;

/// Test cases kept from one report.

pub const MAX_CASES: usize = 50_000;

/// A failure's message, and its output, are cut to these many bytes.

const MAX_MESSAGE: usize = 2 << 10;

const MAX_OUTPUT: usize = 16 << 10;

/// One test case as the report gives it.

#[derive(Clone, Debug, PartialEq, Eq, sqlx::FromRow)]

pub struct Case {

    /// The enclosing `<testsuite>`'s name, `/`-joined when suites nest.

    pub suite: String,

    pub classname: String,

    pub name: String,

    /// `passed`, `failed`, `error` or `skipped`.

    pub status: String,

    pub time_ms: i64,

    /// The failure's, error's or skip's `message`, else its first line.

    pub message: String,

    /// The failure's or error's text, then the case's stdout and stderr;

    /// empty for a passing case.

    pub output: String,

}

impl Case {

    /// `classname::name`, or the name alone.

    pub fn title(&self) -> String {

        if self.classname.is_empty() {

            self.name.clone()

        } else {

            format!("{}::{}", self.classname, self.name)

        }

    }

}

/// Cut `s` to at most `max` bytes on a character boundary, saying so.

fn cut(mut s: String, max: usize) -> String {

    if s.len() > max {

        let mut i = max;

        while !s.is_char_boundary(i) {

            i -= 1;

        }

        s.truncate(i);

        s.push_str("\n… (cut)");

    }

    s

}

fn attr(e: &BytesStart, name: &str) -> String {

    e.try_get_attribute(name)

        .ok()

        .flatten()

        .and_then(|a| a.normalized_value(XmlVersion::Implicit1_0).ok().map(|v| v.into_owned()))

        .unwrap_or_default()

}

/// Seconds as JUnit writes them (`1.25`, sometimes `1,250.5`) to ms.

fn secs_ms(s: &str) -> i64 {

    s.replace(',', "").trim().parse::<f64>().map(|v| (v * 1000.0).round().max(0.0) as i64).unwrap_or(0)

}

/// What the parser is collecting text for.

#[derive(PartialEq)]

enum Collect {

    Nothing,

    Detail,

    Out,

}

/// Parse a JUnit XML report: every `<testcase>`, in order, wherever its

/// `<testsuite>` nests. Counts come from the cases, not the attributes.

pub fn parse(text: &str) -> Result<Vec<Case>, String> {

    let mut r = Reader::from_str(text);

    r.config_mut().trim_text(false);

    let mut suites: Vec<String> = Vec::new();

    let mut cases = Vec::new();

    let mut case: Option<Case> = None;

    let mut detail = String::new();

    let mut out = String::new();

    let mut into = Collect::Nothing;

    let mut seen_root = false;

    loop {

        let ev = r.read_event().map_err(|e| format!("not JUnit XML: {e} (at byte {})", r.error_position()))?;

        match ev {

            Event::Start(ref e) | Event::Empty(ref e)

                if e.local_name().as_ref() == "testsuites" || e.local_name().as_ref() == "testsuite" =>

            {

                seen_root = true;

                if e.local_name().as_ref() == "testsuite" {

                    suites.push(attr(e, "name"));

                }

                // An empty `<testsuite/>` has no end event.

                if matches!(ev, Event::Empty(_)) && e.local_name().as_ref() == "testsuite" {

                    suites.pop();

                }

            }

            Event::Start(ref e) | Event::Empty(ref e) if e.local_name().as_ref() == "testcase" => {

                seen_root = true;

                let c = Case {

                    suite: suites.iter().filter(|s| !s.is_empty()).cloned().collect::<Vec<_>>().join("/"),

                    classname: attr(e, "classname"),

                    name: attr(e, "name"),

                    status: "passed".into(),

                    time_ms: secs_ms(&attr(e, "time")),

                    message: String::new(),

                    output: String::new(),

                };

                if matches!(ev, Event::Empty(_)) {

                    cases.push(c);

                } else {

                    case = Some(c);

                    detail.clear();

                    out.clear();

                }

            }

            Event::Start(ref e) | Event::Empty(ref e) if case.is_some() => {

                let tag = e.local_name().as_ref().to_string();

                let c = case.as_mut().unwrap();

                let status = match tag.as_str() {

                    "failure" | "flakyFailure" if c.status == "passed" => Some("failed"),

                    "error" | "flakyError" if c.status == "passed" => Some("error"),

                    "skipped" if c.status == "passed" => Some("skipped"),

                    _ => None,

                };

                // nextest's flaky retries passed in the end: the case passed.

                let flaky = tag.starts_with("flaky") || tag.starts_with("rerun");

                if let Some(s) = status.filter(|_| !flaky) {

                    c.status = s.into();

                    c.message = attr(e, "message");

                    if matches!(ev, Event::Start(_)) {

                        into = Collect::Detail;

                    }

                } else if matches!(ev, Event::Start(_)) && (tag == "system-out" || tag == "system-err") {

                    into = Collect::Out;

                }

            }

            Event::Text(t) => push_text(&mut into, &mut detail, &mut out, &t.xml10_content()),

            Event::CData(t) => push_text(&mut into, &mut detail, &mut out, &t.xml10_content()),

            Event::GeneralRef(g) => {

                let s = match g.resolve_char_ref() {

                    Ok(Some(ch)) => ch.to_string(),

                    _ => {

                        let name = g.xml10_content();

                        quick_xml::escape::resolve_predefined_entity(&name).map(str::to_string).unwrap_or_else(|| format!("&{name};"))

                    }

                };

                push_text(&mut into, &mut detail, &mut out, &s);

            }

            Event::End(e) => match e.local_name().as_ref() {

                "testsuite" => {

                    suites.pop();

                }

                "testcase" => {

                    if let Some(mut c) = case.take() {

                        if c.status != "passed" {

                            if c.message.is_empty() {

                                c.message = detail.lines().map(str::trim).find(|l| !l.is_empty()).unwrap_or("").to_string();

                            }

                            let mut o = detail.trim().to_string();

                            if !out.trim().is_empty() && c.status != "skipped" {

                                if !o.is_empty() {

                                    o.push_str("\n\n");

                                }

                                o.push_str(out.trim());

                            }

                            c.output = cut(o, MAX_OUTPUT);

                        }

                        c.message = cut(c.message, MAX_MESSAGE);

                        cases.push(c);

                        if cases.len() > MAX_CASES {

                            return Err(format!("more than {MAX_CASES} test cases"));

                        }

                    }

                    into = Collect::Nothing;

                }

                "failure" | "error" | "skipped" | "system-out" | "system-err" | "flakyFailure" | "flakyError" | "rerunFailure"

                | "rerunError" => into = Collect::Nothing,

                _ => {}

            },

            Event::Eof => break,

            _ => {}

        }

    }

    if !seen_root {

        return Err("not JUnit XML: no <testsuites>, <testsuite> or <testcase>".into());

    }

    Ok(cases)

}

fn push_text(into: &mut Collect, detail: &mut String, out: &mut String, s: &str) {

    let (buf, max) = match into {

        Collect::Detail => (detail, MAX_OUTPUT),

        Collect::Out => (out, MAX_OUTPUT),

        Collect::Nothing => return,

    };

    // Keep a little past the cut so `cut` can say it cut.

    if buf.len() <= max {

        buf.push_str(s);

    }

}

/// A run's kept report.

#[derive(Clone, Debug, sqlx::FromRow)]

pub struct Report {

    pub id: i64,

    pub repo_id: i64,

    pub run_id: i64,

    pub job_id: i64,

    pub job: String,

    pub run_number: i64,

    pub bookmark: String,

    pub state: String,

    pub blob: String,

    pub tests: i64,

    pub passed: i64,

    pub failed: i64,

    pub errors: i64,

    pub skipped: i64,

    pub time_ms: i64,

    pub created_ms: i64,

}

impl Report {

    /// "630 tests: 2 failed, 3 skipped", or "630 passed".

    pub fn summary(&self) -> String {

        let mut bad = vec![];

        if self.failed > 0 {

            bad.push(format!("{} failed", self.failed));

        }

        if self.errors > 0 {

            bad.push(format!("{} errored", self.errors));

        }

        if self.skipped > 0 {

            bad.push(format!("{} skipped", self.skipped));

        }

        let tests = if self.tests == 1 { "1 test".to_string() } else { format!("{} tests", self.tests) };

        if bad.is_empty() {

            format!("{tests}, all passed")

        } else {

            format!("{tests}: {}", bad.join(", "))

        }

    }

    pub fn broken(&self) -> i64 {

        self.failed + self.errors

    }

}

/// What a run's report is kept with.

pub struct Origin<'a> {

    pub repo_id: i64,

    pub run_id: i64,

    pub run_number: i64,

    pub job_id: i64,

    pub job: &'a str,

    pub bookmark: &'a str,

    pub state: &'a str,

    pub blob: &'a str,

}

/// Keep a parsed report, replacing an earlier one of the same job.

pub async fn keep(app: &Shared, o: &Origin<'_>, cases: &[Case]) -> sqlx::Result<i64> {

    let count = |s: &str| cases.iter().filter(|c| c.status == s).count() as i64;

    let mut tx = crate::db::begin_write(&app.db).await?;

    sqlx::query("DELETE FROM test_reports WHERE job_id = ?").bind(o.job_id).execute(&mut *tx).await?;

    let id: i64 = sqlx::query_scalar(

        "INSERT INTO test_reports (repo_id, run_id, job_id, job, run_number, bookmark, state, blob, \

         tests, passed, failed, errors, skipped, time_ms, created_ms) \

         VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?) RETURNING id",

    )

    .bind(o.repo_id)

    .bind(o.run_id)

    .bind(o.job_id)

    .bind(o.job)

    .bind(o.run_number)

    .bind(o.bookmark)

    .bind(o.state)

    .bind(o.blob)

    .bind(cases.len() as i64)

    .bind(count("passed"))

    .bind(count("failed"))

    .bind(count("error"))

    .bind(count("skipped"))

    .bind(cases.iter().map(|c| c.time_ms).sum::<i64>())

    .bind(now_ms())

    .fetch_one(&mut *tx)

    .await?;

    for (i, c) in cases.iter().enumerate() {

        sqlx::query(

            "INSERT INTO test_cases (report_id, idx, suite, classname, name, status, time_ms, message, output) \

             VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",

        )

        .bind(id)

        .bind(i as i64)

        .bind(&c.suite)

        .bind(&c.classname)

        .bind(&c.name)

        .bind(&c.status)

        .bind(c.time_ms)

        .bind(&c.message)

        .bind(&c.output)

        .execute(&mut *tx)

        .await?;

    }

    tx.commit().await?;

    mark(app, o.repo_id);

    Ok(id)

}

/// A run's reports, one per job that kept one.

pub async fn for_run(pool: &Pool, run_id: i64) -> sqlx::Result<Vec<Report>> {

    sqlx::query_as("SELECT * FROM test_reports WHERE run_id = ? ORDER BY job").bind(run_id).fetch_all(pool).await

}

/// A report's cases in report order, all or only those with a status

/// among `statuses`.

pub async fn cases(pool: &Pool, report_id: i64, statuses: &[&str], limit: i64) -> sqlx::Result<Vec<Case>> {

    let filter = if statuses.is_empty() {

        String::new()

    } else {

        format!(" AND status IN ({})", statuses.iter().map(|_| "?").collect::<Vec<_>>().join(", "))

    };

    let sql = format!(

        "SELECT suite, classname, name, status, time_ms, message, output FROM test_cases WHERE report_id = ?{filter} ORDER BY idx LIMIT ?"

    );

    let mut q = sqlx::query_as(&sql).bind(report_id);

    for s in statuses {

        q = q.bind(*s);

    }

    q.bind(limit).fetch_all(pool).await

}

/// A report's failed and errored cases, in report order.

pub async fn failures(pool: &Pool, report_id: i64, limit: i64) -> sqlx::Result<Vec<Case>> {

    sqlx::query_as(

        "SELECT suite, classname, name, status, time_ms, message, output FROM test_cases \

         WHERE report_id = ? AND status IN ('failed', 'error') ORDER BY idx LIMIT ?",

    )

    .bind(report_id)

    .bind(limit)

    .fetch_all(pool)

    .await

}

/// Which repositories have test reports, for the Tests tab; keyed by data

/// directory as well, for tests running several hubs in one process.

static HAS: std::sync::Mutex<Option<std::collections::HashSet<String>>> = std::sync::Mutex::new(None);

fn key(app: &Shared, repo_id: i64) -> String {

    format!("{}#{repo_id}", app.repos.root().display())

}

pub fn has_reports(app: &Shared, repo_id: i64) -> bool {

    HAS.lock().unwrap_or_else(|p| p.into_inner()).as_ref().is_some_and(|s| s.contains(&key(app, repo_id)))

}

fn mark(app: &Shared, repo_id: i64) {

    HAS.lock().unwrap_or_else(|p| p.into_inner()).get_or_insert_with(Default::default).insert(key(app, repo_id));

}

/// At start: which repositories have reports.

pub async fn load(app: &Shared) -> sqlx::Result<()> {

    let ids: Vec<(i64,)> = sqlx::query_as("SELECT DISTINCT repo_id FROM test_reports").fetch_all(&app.db).await?;

    for (id,) in ids {

        mark(app, id);

    }

    Ok(())

}

pub async fn report(pool: &Pool, repo_id: i64, id: i64) -> sqlx::Result<Option<Report>> {

    sqlx::query_as("SELECT * FROM test_reports WHERE id = ? AND repo_id = ?").bind(id).bind(repo_id).fetch_optional(pool).await

}

/// The newest reports on a bookmark, newest first.

pub async fn on_bookmark(pool: &Pool, repo_id: i64, bookmark: &str, limit: i64) -> sqlx::Result<Vec<Report>> {

    sqlx::query_as("SELECT * FROM test_reports WHERE repo_id = ? AND bookmark = ? ORDER BY created_ms DESC, id DESC LIMIT ?")

        .bind(repo_id)

        .bind(bookmark)

        .bind(limit)

        .fetch_all(pool)

        .await

}

/// Bookmarks with reports, the most recently reported first.

pub async fn bookmarks(pool: &Pool, repo_id: i64) -> sqlx::Result<Vec<String>> {

    let rows: Vec<(String,)> =

        sqlx::query_as("SELECT bookmark FROM test_reports WHERE repo_id = ? GROUP BY bookmark ORDER BY MAX(created_ms) DESC")

            .bind(repo_id)

            .fetch_all(pool)

            .await?;

    Ok(rows.into_iter().map(|(b,)| b).collect())

}

pub async fn first_ms(pool: &Pool, repo_id: i64) -> sqlx::Result<Option<i64>> {

    sqlx::query_scalar("SELECT MIN(created_ms) FROM test_reports WHERE repo_id = ?").bind(repo_id).fetch_one(pool).await

}

/// A report's slowest cases.

pub async fn slowest(pool: &Pool, report_id: i64, limit: i64) -> sqlx::Result<Vec<Case>> {

    sqlx::query_as(

        "SELECT suite, classname, name, status, time_ms, message, '' AS output FROM test_cases \

         WHERE report_id = ? AND status != 'skipped' ORDER BY time_ms DESC, idx LIMIT ?",

    )

    .bind(report_id)

    .bind(limit)

    .fetch_all(pool)

    .await

}

/// Every case of a report as `(classname, name) -> status`.

pub async fn statuses(pool: &Pool, report_id: i64) -> sqlx::Result<std::collections::HashMap<(String, String), String>> {

    let rows: Vec<(String, String, String)> =

        sqlx::query_as("SELECT classname, name, status FROM test_cases WHERE report_id = ?").bind(report_id).fetch_all(pool).await?;

    Ok(rows.into_iter().map(|(c, n, s)| ((c, n), s)).collect())

}

/// A test that both failed and passed across some reports.

#[derive(Clone, Debug, sqlx::FromRow)]

pub struct Flaky {

    pub classname: String,

    pub name: String,

    pub failed: i64,

    pub passed: i64,

}

impl Flaky {

    pub fn title(&self) -> String {

        if self.classname.is_empty() {

            self.name.clone()

        } else {

            format!("{}::{}", self.classname, self.name)

        }

    }

}

/// Tests that failed in some of `reports` and passed in others, the most

/// often failing first: flaky, or broken and fixed again.

pub async fn flaky(pool: &Pool, reports: &[i64], limit: i64) -> sqlx::Result<Vec<Flaky>> {

    if reports.is_empty() {

        return Ok(vec![]);

    }

    let ids = reports.iter().map(|i| i.to_string()).collect::<Vec<_>>().join(",");

    sqlx::query_as(&format!(

        "SELECT classname, name, SUM(status IN ('failed', 'error')) AS failed, SUM(status = 'passed') AS passed \

         FROM test_cases WHERE report_id IN ({ids}) GROUP BY classname, name \

         HAVING failed > 0 AND passed > 0 ORDER BY failed DESC, classname, name LIMIT ?"

    ))

    .bind(limit)

    .fetch_all(pool)

    .await

}

#[cfg(test)]

mod tests {

    use super::*;

    #[test]

    fn pytest_nextest_and_jest_shapes() {

        // pytest: one suite, a failure with a message and text, a skip, an error.

        let py = r#"<?xml version="1.0" encoding="utf-8"?>

<testsuites><testsuite name="pytest" errors="1" failures="1" skipped="1" tests="4" time="3.5">

<testcase classname="tests.test_a" name="test_ok" time="0.010"/>

<testcase classname="tests.test_a" name="test_bad" time="1.25"><failure message="assert 1 == 2">def test_bad():

>       assert 1 == 2

E       assert 1 == 2</failure><system-out>printed &amp; kept</system-out></testcase>

<testcase classname="tests.test_b" name="test_skip" time="0"><skipped type="pytest.skip" message="no emulator"/></testcase>

<testcase classname="tests.test_b" name="test_err" time="0.5"><error message="fixture failed"><![CDATA[Traceback <here>]]></error></testcase>

</testsuite></testsuites>"#;

        let c = parse(py).unwrap();

        let st: Vec<(&str, &str)> = c.iter().map(|c| (c.name.as_str(), c.status.as_str())).collect();

        assert_eq!(st, [("test_ok", "passed"), ("test_bad", "failed"), ("test_skip", "skipped"), ("test_err", "error")]);

        assert_eq!(c[1].message, "assert 1 == 2");