letterbox/server/src/lib.rs

pub mod error;
pub mod graphql;
pub mod newsreader;
pub mod nm;

use css_inline::{CSSInliner, InlineError, InlineOptions};
use linkify::{LinkFinder, LinkKind};
use log::error;
use lol_html::{element, errors::RewritingError, rewrite_str, RewriteStrSettings};
use maplit::{hashmap, hashset};
use thiserror::Error;

#[derive(Error, Debug)]
pub enum SanitizeError {
    #[error("lol-html rewrite error")]
    RewritingError(#[from] RewritingError),
    #[error("css inline error")]
    InlineError(#[from] InlineError),
}

pub fn linkify_html(text: &str) -> String {
    let mut finder = LinkFinder::new();
    let finder = finder.url_must_have_scheme(false).kinds(&[LinkKind::Url]);
    let mut parts = Vec::new();
    for span in finder.spans(text) {
        // TODO(wathiede): use Cow<str>?
        match span.kind() {
            // Text as-is
            None => parts.push(span.as_str().to_string()),
            // Wrap in anchor tag
            Some(LinkKind::Url) => {
                let text = span.as_str();
                let schema = if text.starts_with("http") {
                    ""
                } else {
                    "http://"
                };
                let a = format!(r#"<a href="{schema}{0}">{0}</a>"#, text);
                parts.push(a);
            }
            _ => todo!("unhandled kind: {:?}", span.kind().unwrap()),
        }
    }
    parts.join("")
}

// html contains the content to be cleaned, and cid_prefix is used to resolve mixed part image
// referrences
pub fn sanitize_html(html: &str, cid_prefix: &str) -> Result<String, SanitizeError> {
    let element_content_handlers = vec![
        // Open links in new tab
        element!("a[href]", |el| {
            el.set_attribute("target", "_blank").unwrap();

            Ok(())
        }),
        // Replace mixed part CID images with URL
        element!("img[src]", |el| {
            let src = el
                .get_attribute("src")
                .expect("src was required")
                .replace("cid:", cid_prefix);

            el.set_attribute("src", &src)?;

            Ok(())
        }),
        // Only secure image URLs
        element!("img[src]", |el| {
            let src = el
                .get_attribute("src")
                .expect("src was required")
                .replace("http:", "https:");

            el.set_attribute("src", &src)?;

            Ok(())
        }),
    ];

    let inline_opts = InlineOptions {
        inline_style_tags: true,
        keep_style_tags: false,
        keep_link_tags: false,
        base_url: None,
        load_remote_stylesheets: false,
        extra_css: None,
        preallocate_node_capacity: 32,
        ..InlineOptions::default()
    };

    let inlined_html = match CSSInliner::new(inline_opts).inline(&html) {
        Ok(inlined_html) => inlined_html,
        Err(err) => {
            error!("failed to inline CSS: {err}");
            html.to_string()
        }
    };
    // Default's don't allow style, but we want to preserve that.
    // TODO: remove 'class' if rendering mails moves to a two phase process where abstract message
    // types are collected, santized, and then grouped together as one big HTML doc
    let attributes = hashset![
        "align", "bgcolor", "class", "color", "height", "lang", "title", "width", "style",
    ];

    let tags = hashset![
        "a",
        "abbr",
        "acronym",
        "area",
        "article",
        "aside",
        "b",
        "bdi",
        "bdo",
        "blockquote",
        "br",
        "caption",
        "center",
        "cite",
        "code",
        "col",
        "colgroup",
        "data",
        "dd",
        "del",
        "details",
        "dfn",
        "div",
        "dl",
        "dt",
        "em",
        "figcaption",
        "figure",
        "footer",
        "h1",
        "h2",
        "h3",
        "h4",
        "h5",
        "h6",
        "header",
        "hgroup",
        "hr",
        "i",
        "img",
        "ins",
        "kbd",
        "kbd",
        "li",
        "map",
        "mark",
        "nav",
        "ol",
        "p",
        "pre",
        "q",
        "rp",
        "rt",
        "rtc",
        "ruby",
        "s",
        "samp",
        "small",
        "span",
        "strike",
        "strong",
        "sub",
        "summary",
        "sup",
        "table",
        "tbody",
        "td",
        "th",
        "thead",
        "time",
        "title", // wathiede
        "tr",
        "tt",
        "u",
        "ul",
        "var",
        "wbr",
    ];
    let tag_attributes = hashmap![
        "a" => hashset![
            "href", "hreflang", "target",
        ],
        "bdo" => hashset![
            "dir"
        ],
        "blockquote" => hashset![
            "cite"
        ],
        "col" => hashset![
            "align", "char", "charoff", "span"
        ],
        "colgroup" => hashset![
            "align", "char", "charoff", "span"
        ],
        "del" => hashset![
            "cite", "datetime"
        ],
        "hr" => hashset![
            "align", "size", "width"
        ],
        "img" => hashset![
            "align", "alt", "height", "src", "width"
        ],
        "ins" => hashset![
            "cite", "datetime"
        ],
        "ol" => hashset![
            "start"
        ],
        "q" => hashset![
            "cite"
        ],
        "table" => hashset![
            "align", "border", "cellpadding", "cellspacing", "char", "charoff", "summary",
        ],
        "tbody" => hashset![
            "align", "char", "charoff"
        ],
        "td" => hashset![
            "align", "char", "charoff", "colspan", "headers", "rowspan"
        ],
        "tfoot" => hashset![
            "align", "char", "charoff"
        ],
        "th" => hashset![
            "align", "char", "charoff", "colspan", "headers", "rowspan", "scope"
        ],
        "thead" => hashset![
            "align", "char", "charoff"
        ],
        "tr" => hashset![
            "align", "char", "charoff"
        ],
    ];

    let rewritten_html = rewrite_str(
        &inlined_html,
        RewriteStrSettings {
            element_content_handlers,
            ..RewriteStrSettings::default()
        },
    )?;
    let clean_html = ammonia::Builder::default()
        .tags(tags)
        .tag_attributes(tag_attributes)
        .generic_attributes(attributes)
        .clean(&rewritten_html)
        .to_string();

    Ok(clean_html)
}