13. テストで固める — トークン列を丸ごと比べ、境界と壊れた入力を押さえる

// src/html/tokenizer.rs の末尾のテスト
#[cfg(test)]
mod tests {
    use super::*;

    fn tokens(html: &str) -> Vec<HtmlToken> {
        HtmlTokenizer::new(html).collect()
    }

    fn start(tag: &str, attrs: &[(&str, &str)]) -> HtmlToken {
        HtmlToken::StartTag {
            tag: tag.to_string(),
            self_closing: false,
            attributes: attrs.iter().map(|(n, v)| Attribute::new(n, v)).collect(),
        }
    }

    fn end(tag: &str) -> HtmlToken {
        HtmlToken::EndTag { tag: tag.to_string() }
    }

    fn chars(s: &str) -> Vec<HtmlToken> {
        s.chars().map(HtmlToken::Char).collect()
    }

    #[test]
    fn empty_input_gives_only_eof() {
        assert_eq!(tokens(""), vec![HtmlToken::Eof]);
    }

    #[test]
    fn start_and_end_tag() {
        let mut expected = vec![start("p", &[])];
        expected.extend(chars("hi"));
        expected.push(end("p"));
        expected.push(HtmlToken::Eof);
        assert_eq!(tokens("<p>hi</p>"), expected);
    }

    #[test]
    fn tag_names_are_lowercased() {
        assert_eq!(tokens("<DIV></Div>"), vec![start("div", &[]), end("div"), HtmlToken::Eof]);
    }

    #[test]
    fn attributes_in_three_quote_styles() {
        let t = tokens(r#"<a href="/x" title='t' target=_blank>"#);
        assert_eq!(
            t[0],
            start("a", &[("href", "/x"), ("title", "t"), ("target", "_blank")])
        );
    }

    #[test]
    fn attribute_without_value_and_self_closing() {
        let t = tokens("<input disabled required/>");
        assert_eq!(
            t[0],
            HtmlToken::StartTag {
                tag: "input".into(),
                self_closing: true,
                attributes: vec![Attribute::new("disabled", ""), Attribute::new("required", "")],
            }
        );
    }

    #[test]
    fn lone_less_than_is_a_character() {
        let t = tokens("a < b");
        assert_eq!(t, {
            let mut v = chars("a < b");
            v.push(HtmlToken::Eof);
            v
        });
    }

    #[test]
    fn script_content_is_not_parsed_as_tags() {
        let t = tokens("<script>if (a < b) { x = '</div>'; }</script>");
        let body: String = t[1..t.len() - 2]
            .iter()
            .map(|t| match t {
                HtmlToken::Char(c) => *c,
                other => panic!("script の中にタグが出た: {other:?}"),
            })
            .collect();
        assert_eq!(body, "if (a < b) { x = '</div>'; }");
        assert_eq!(t[t.len() - 2], end("script"));
    }

    #[test]
    fn doctype_and_comment_are_skipped() {
        assert_eq!(
            tokens("<!DOCTYPE html><!-- c --><p>"),
            vec![start("p", &[]), HtmlToken::Eof]
        );
    }
}