artifactview/tests/tests.rs

use std::{
    io::Read,
    path::{Path, PathBuf},
    time::{Duration, SystemTime},
};

use axum_test::{TestRequest, TestResponse, TestServer};
use headers::HeaderMapExt;
use http::{header, HeaderName, HeaderValue, StatusCode};
use once_cell::sync::Lazy;
use path_macro::path;
use rstest::{fixture, rstest};

use artifactview::{App, AppState, Config, ConfigData};
use scraper::{selectable::Selectable, ElementRef, Html, Selector};
use temp_testdir::TempDir;

static TESTFILES: Lazy<PathBuf> =
    Lazy::new(|| path!(env!("CARGO_MANIFEST_DIR") / "tests" / "testfiles"));

static SITEDIR: Lazy<PathBuf> = Lazy::new(|| {
    let sitedir = path!(*TESTFILES / "sites_data");
    if !sitedir.is_dir() {
        std::process::Command::new(path!(*TESTFILES / "sites" / "make_zip.sh"))
            .output()
            .unwrap();
    }
    sitedir
});

static CACHEDIR: Lazy<TempDir> = Lazy::new(|| {
    let td = TempDir::default();
    setup_cache_dir(&td);
    td
});

const S1: &str = "codeberg-org--thetadev--artifactview-test--1-1";
const S2: &str = "codeberg-org--thetadev--artifactview-test--1-2";
const S3: &str = "codeberg-org--thetadev--artifactview-test--1-3";

fn setup_cache_dir(dir: &Path) {
    for entry in std::fs::read_dir(SITEDIR.as_path()).unwrap() {
        let entry = entry.unwrap();
        if entry.file_type().unwrap().is_file() {
            std::fs::copy(entry.path(), path!(dir / entry.file_name())).unwrap();
        }
    }
}

struct TestAv {
    server: TestServer,
}

#[fixture]
fn server() -> TestAv {
    let router = App::router(AppState::from_cfg(
        Config::from_data(ConfigData {
            cache_dir: CACHEDIR.to_path_buf(),
            no_https: true,
            real_ip_header: Some("x-forwarded-for".to_string()),
            ..Default::default()
        })
        .unwrap(),
    ));

    let mut server = TestServer::new(router).unwrap();

    server.add_header(
        HeaderName::from_static("x-forwarded-for"),
        HeaderValue::from_static("127.0.0.1"),
    );

    TestAv { server }
}

impl TestAv {
    fn get(&self, subdomain: &str, path: &str) -> TestRequest {
        self.server.get(path).add_header(
            header::HOST,
            if subdomain.is_empty() {
                HeaderValue::from_static("localhost:3000")
            } else {
                HeaderValue::from_str(&format!("{subdomain}.localhost:3000")).unwrap()
            },
        )
    }

    async fn get_html(&self, subdomain: &str, path: &str) -> Html {
        let resp = self.get(subdomain, path).await;
        resp.assert_status_ok();
        assert_eq!(resp.header(header::CONTENT_TYPE), "text/html");
        scraper::Html::parse_document(&resp.text())
    }
}

#[derive(Debug)]
struct FileEntry {
    name: String,
    size: String,
    crc32: String,
}

#[derive(Debug, PartialEq, Eq)]
struct Link {
    text: String,
    href: String,
    selected: bool,
}

fn text_from_elm(elm: &ElementRef<'_>) -> String {
    elm.text()
        .fold(String::new(), |acc, s| acc + s)
        .trim()
        .to_owned()
}

fn parse_link(elm: &ElementRef<'_>) -> Link {
    assert_eq!(elm.value().name(), "a");
    Link {
        text: text_from_elm(elm),
        href: elm.attr("href").expect("href").to_owned(),
        selected: elm
            .value()
            .has_class("selected", scraper::CaseSensitivity::CaseSensitive),
    }
}

#[track_caller]
fn assert_cache_immutable(resp: &TestResponse) {
    assert_eq!(
        resp.header(header::CACHE_CONTROL),
        "max-age=31536000,public,immutable"
    );
}

#[rstest]
#[tokio::test]
async fn homepage(server: TestAv) {
    let doc = server.get_html("", "/").await;
    let elm = doc
        .select(&Selector::parse("title").unwrap())
        .next()
        .unwrap();
    assert_eq!(text_from_elm(&elm), "Artifactview");
}

#[rstest]
#[case(S1, "/example.rs", true, "example.rs", "text/x-rust")]
#[case(S1, "/sites/", true, "index.html", "text/html")]
#[case(S1, "/foo.txt", false, "", "text/html")]
#[case(S1, "/.well-known/test.txt", false, "", "text/html")]
// 404 fallback
#[case(S2, "/foo.txt", false, "404.html", "text/html")]
// SPA
#[case(S3, "/", true, "index.html", "text/html")]
#[case(S3, "/foo.txt", true, "200.html", "text/html")]
#[tokio::test]
async fn get_file(
    server: TestAv,
    #[case] sd: &str,
    #[case] path: &str,
    #[case] found: bool,
    #[case] exp_path: &str,
    #[case] mime: &str,
) {
    let resp = server.get(sd, path).await;
    assert_eq!(resp.header(header::CONTENT_TYPE), mime);

    if found {
        resp.assert_status_ok();
        assert_cache_immutable(&resp);
    } else {
        resp.assert_status_not_found();
    }

    if !exp_path.is_empty() {
        let expect = std::fs::read_to_string(path!(*TESTFILES / "sites" / exp_path)).unwrap();
        assert_eq!(resp.text(), expect);
    }
}

#[rstest]
#[tokio::test]
async fn robots_txt(server: TestAv) {
    let resp1 = server.get("", "/robots.txt").await;
    let resp2 = server.get(S1, "/robots.txt").await;

    assert!(resp1.text().contains("\nUser-agent: *\nDisallow: /\n"));
    assert!(resp2.text().contains("\nUser-agent: *\nDisallow: /\n"));
}

#[rstest]
#[tokio::test]
async fn stylesheet(server: TestAv) {
    let html = server.get_html(S1, "/README.md?viewer=1").await;
    for path in html
        .select(&Selector::parse("link[rel=\"stylesheet\"]").unwrap())
        .map(|elm| {
            elm.attr("href")
                .expect("href")
                .strip_prefix("http://localhost:3000")
                .expect("localhost url")
        })
    {
        let resp = server.get("", path).await;
        resp.assert_status_ok();
        assert_eq!(resp.header(header::CONTENT_TYPE), "text/css");
        assert_cache_immutable(&resp);

        // Remove running number from stylesheet path
        let fname = path
            .strip_prefix('/')
            .unwrap()
            .strip_suffix(".css")
            .unwrap()
            .trim_end_matches(['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']);
        let expect = std::fs::read_to_string(path!(
            env!("CARGO_MANIFEST_DIR") / "resources" / format!("{fname}.css")
        ))
        .unwrap();
        assert_eq!(resp.text(), expect);
    }
}

fn parse_listing(doc: &Html) -> Vec<FileEntry> {
    let sel_col = Selector::parse("td").unwrap();
    doc.select(&Selector::parse(".file").unwrap())
        .map(|elm| {
            let icn = elm
                .select(&Selector::parse("svg > use").unwrap())
                .next()
                .expect("icon");
            let (k, icn_id) = icn.value().attrs().next().expect("icon id");
            assert_eq!(k, "href");
            assert!(matches!(icn_id, "#file" | "#folder"));

            let mut parts = elm.select(&sel_col).map(|elm| text_from_elm(&elm));

            let name = parts.next().expect("name");
            let folder = icn_id == "#folder";
            assert_eq!(name.ends_with('/'), folder);

            FileEntry {
                name,
                crc32: parts.next().expect("crc32"),
                size: parts.next().expect("size"),
            }
        })
        .collect()
}

#[rstest]
#[case("/", &[
    ".well-known/",
    "junit/",
    "sites/",
    "README.md",
    "example.rs",
    "robots.txt"
])]
#[case("/?C=N&O=D", &[
    "sites/",
    "junit/",
    ".well-known/",
    "robots.txt",
    "example.rs",
    "README.md",
])]
#[tokio::test]
async fn listing(server: TestAv, #[case] path: &str, #[case] expect: &[&str]) {
    let doc = server.get_html(S1, path).await;
    let files = parse_listing(&doc);
    let file_names = files.iter().map(|f| f.name.to_owned()).collect::<Vec<_>>();

    assert_eq!(file_names, expect);

    let example_rs = files
        .iter()
        .find(|f| f.name == "example.rs")
        .expect("example.rs");
    assert_eq!(example_rs.size, "406 B");
    assert_eq!(example_rs.crc32, "2013120c");
}

#[rstest]
// JUnit
#[case("/junit/hello.junit.xml?viewer=1", "junit", &[("junit", "JUnit"), ("code", "Code")])]
#[case("/junit/hello.junit.xml?viewer=junit", "junit", &[("junit", "JUnit"), ("code", "Code")])]
#[case("/junit/hello.junit.xml?viewer=code", "code", &[("junit", "JUnit"), ("code", "Code")])]
#[case("/junit/hello.junit.xml?viewer=md", "", &[])]
#[case("/example.rs?viewer=1", "code", &[("code", "Code")])]
#[case("/README.md?viewer=1", "md", &[("md", "Markdown"), ("code", "Code")])]
#[tokio::test]
async fn viewer(
    server: TestAv,
    #[case] path: &str,
    #[case] vid: &str,
    #[case] exp_viewers: &[(&str, &str)],
) {
    if vid.is_empty() {
        // Raw file
        let resp = server.get(S1, path).await;
        resp.assert_status_ok();
        assert_eq!(resp.header(header::CONTENT_TYPE), "text/xml");
        assert_cache_immutable(&resp);
    } else {
        let doc = server.get_html(S1, path).await;
        let viewers = doc
            .select(&Selector::parse("#viewers a").unwrap())
            .map(|elm| parse_link(&elm))
            .filter(|ln| ln.text != "Raw")
            .collect::<Vec<_>>();

        let viewer_names = viewers
            .iter()
            .map(|v| {
                (
                    v.href.strip_prefix("?viewer=").expect("link prefix"),
                    v.text.as_str(),
                )
            })
            .collect::<Vec<_>>();
        assert_eq!(viewer_names, exp_viewers);

        // Selected viewer
        let mut sel_iter = viewers.iter().filter(|ln| ln.selected);
        let selected = sel_iter.next().expect("selected");
        assert_eq!(selected.href, format!("?viewer={vid}"));
        assert!(sel_iter.next().is_none());
    }
}

#[rstest]
#[tokio::test]
async fn if_modified(server: TestAv) {
    let resp = server.get(S1, "/README.md").await;
    let lastmod = SystemTime::from(
        resp.headers()
            .typed_get::<headers::LastModified>()
            .expect("last modified"),
    );
    let bef_lastmod = lastmod - Duration::from_secs(1);

    // if-modified-since
    let resp = server
        .get(S1, "/README.md")
        .add_header(
            header::IF_MODIFIED_SINCE,
            HeaderValue::from_str(&httpdate::HttpDate::from(lastmod).to_string()).unwrap(),
        )
        .await;
    resp.assert_status(StatusCode::NOT_MODIFIED);

    let resp = server
        .get(S1, "/README.md")
        .add_header(
            header::IF_MODIFIED_SINCE,
            HeaderValue::from_str(&httpdate::HttpDate::from(bef_lastmod).to_string()).unwrap(),
        )
        .await;
    resp.assert_status_ok();

    // if-unmodified-since
    let resp = server
        .get(S1, "/README.md")
        .add_header(
            header::IF_UNMODIFIED_SINCE,
            HeaderValue::from_str(&httpdate::HttpDate::from(lastmod).to_string()).unwrap(),
        )
        .await;
    resp.assert_status_ok();

    let resp = server
        .get(S1, "/README.md")
        .add_header(
            header::IF_UNMODIFIED_SINCE,
            HeaderValue::from_str(&httpdate::HttpDate::from(bef_lastmod).to_string()).unwrap(),
        )
        .await;
    resp.assert_status(StatusCode::PRECONDITION_FAILED);
}

#[rstest]
#[tokio::test]
async fn range_request(server: TestAv) {
    let resp = server
        .get(S1, "/example.rs")
        .add_header(header::RANGE, HeaderValue::from_static("bytes=10-99"))
        .await;
    resp.assert_status(StatusCode::PARTIAL_CONTENT);
    assert_cache_immutable(&resp);
    assert_eq!(resp.header(header::CONTENT_TYPE), "text/x-rust");
    assert_eq!(resp.header(header::CONTENT_LENGTH), "90");
    assert_eq!(resp.header(header::CONTENT_RANGE), "bytes 10-99/406");

    let expect = std::fs::read_to_string(path!(*TESTFILES / "sites" / "example.rs")).unwrap();
    assert_eq!(resp.text(), &expect[10..100]);
}

#[rstest]
#[tokio::test]
async fn range_request_start(server: TestAv) {
    let resp = server
        .get(S1, "/example.rs")
        .add_header(header::RANGE, HeaderValue::from_static("bytes=100-"))
        .await;
    resp.assert_status(StatusCode::PARTIAL_CONTENT);
    assert_cache_immutable(&resp);
    assert_eq!(resp.header(header::CONTENT_TYPE), "text/x-rust");
    assert_eq!(resp.header(header::CONTENT_LENGTH), "306");
    assert_eq!(resp.header(header::CONTENT_RANGE), "bytes 100-405/406");

    let expect = std::fs::read_to_string(path!(*TESTFILES / "sites" / "example.rs")).unwrap();
    assert_eq!(resp.text(), &expect[100..]);
}

#[rstest]
#[tokio::test]
async fn compressed(server: TestAv) {
    let resp = server
        .get(S1, "/example.rs")
        .add_header(header::ACCEPT_ENCODING, HeaderValue::from_static("gzip"))
        .await;
    let bts = resp.into_bytes().to_vec();
    let mut gz = flate2::read::GzDecoder::new(&bts[..]);

    let mut buf = String::new();
    gz.read_to_string(&mut buf).unwrap();

    let expect = std::fs::read_to_string(path!(*TESTFILES / "sites" / "example.rs")).unwrap();
    assert_eq!(buf, expect);
}

#[rstest]
#[tokio::test]
async fn userscript(server: TestAv) {
    let resp = server.get("", "/artifactview.user.js").await;
    resp.assert_status_ok();
    let script = resp.text();
    assert!(script.starts_with("// ==UserScript==\n"));
}

#[rstest]
#[tokio::test]
/// Redirect user to the directory path if index.html or directory listing is served
async fn index_redirect(server: TestAv) {
    let resp = server.get(S1, "/sites").await;
    resp.assert_status(StatusCode::PERMANENT_REDIRECT);
    assert_eq!(resp.header(header::LOCATION), "/sites/");

    let resp = server.get(S1, "/junit").await;
    resp.assert_status(StatusCode::PERMANENT_REDIRECT);
    assert_eq!(resp.header(header::LOCATION), "/junit/");
}