server: escape RSS feeds that are HTML escaped

server/src/lib.rs

@@ -11,14 +11,49 @@ use maplit::{hashmap, hashset};
 use thiserror::Error;
 use url::Url;
 
+// TODO: figure out how to use Cow
+trait Transformer {
+    fn should_run(&self, input: &str) -> bool;
+    // TODO: should input be something like `html_escape` uses:
+    // <S: ?Sized + AsRef<str>>(text: &S) -> Cow<str>
+    fn transform(&self, input: &str) -> Result<String, TransformError>;
+}
+
+// TODO: how would we make this more generic to allow good implementations of Transformer outside
+// of this module?
 #[derive(Error, Debug)]
-pub enum SanitizeError {
+pub enum TransformError {
     #[error("lol-html rewrite error")]
     RewritingError(#[from] RewritingError),
     #[error("css inline error")]
     InlineError(#[from] InlineError),
 }
 
+struct SanitizeHtml<'a> {
+    cid_prefix: &'a str,
+    base_url: &'a Option<Url>,
+}
+
+impl<'a> Transformer for SanitizeHtml<'a> {
+    fn should_run(&self, _input: &str) -> bool {
+        true
+    }
+    fn transform(&self, input: &str) -> Result<String, TransformError> {
+        Ok(sanitize_html(input, self.cid_prefix, self.base_url)?)
+    }
+}
+
+struct EscapeHtml;
+
+impl Transformer for EscapeHtml {
+    fn should_run(&self, input: &str) -> bool {
+        input.starts_with("&lt")
+    }
+    fn transform(&self, input: &str) -> Result<String, TransformError> {
+        Ok(html_escape::decode_html_entities(input).to_string())
+    }
+}
+
 pub fn linkify_html(text: &str) -> String {
     let mut finder = LinkFinder::new();
     let finder = finder.url_must_have_scheme(false).kinds(&[LinkKind::Url]);
@@ -51,7 +86,7 @@ pub fn sanitize_html(
     html: &str,
     cid_prefix: &str,
     base_url: &Option<Url>,
-) -> Result<String, SanitizeError> {
+) -> Result<String, TransformError> {
     let mut element_content_handlers = vec![
         // Open links in new tab
         element!("a[href]", |el| {
@@ -86,10 +121,7 @@ pub fn sanitize_html(
         element_content_handlers.extend(vec![
             // Make links with relative URLs absolute
             element!("a[href]", |el| {
-                if let Some(Ok(href)) = el.get_attribute("href").map(|href| {
-                    info!("href {href:?}");
-                    base_url.join(&href)
-                }) {
+                if let Some(Ok(href)) = el.get_attribute("href").map(|href| base_url.join(&href)) {
                     el.set_attribute("href", &href.as_str()).unwrap();
                 }
 
@@ -98,7 +130,6 @@ pub fn sanitize_html(
             // Make images with relative srcs absolute
             element!("img[src]", |el| {
                 if let Some(Ok(src)) = el.get_attribute("src").map(|src| base_url.join(&src)) {
-                    info!("src {src:?}");
                     el.set_attribute("src", &src.as_str()).unwrap();
                 }