Add support for image and link URL rewriting

src/html.rs

@@ -864,7 +864,11 @@ impl<'o, 'c: 'o> HtmlFormatter<'o, 'c> {
                         self.output.write_all(b" href=\"")?;
                         let url = nl.url.as_bytes();
                         if self.options.render.unsafe_ || !dangerous_url(url) {
-                            self.escape_href(url)?;
+                            if let Some(rewriter) = &self.options.extension.link_url_rewriter {
+                                self.escape_href(rewriter.to_html(&nl.url).as_bytes())?;
+                            } else {
+                                self.escape_href(url)?;
+                            }
                         }
                         if !nl.title.is_empty() {
                             self.output.write_all(b"\" title=\"")?;
@@ -889,7 +893,11 @@ impl<'o, 'c: 'o> HtmlFormatter<'o, 'c> {
                     self.output.write_all(b" src=\"")?;
                     let url = nl.url.as_bytes();
                     if self.options.render.unsafe_ || !dangerous_url(url) {
-                        self.escape_href(url)?;
+                        if let Some(rewriter) = &self.options.extension.image_url_rewriter {
+                            self.escape_href(rewriter.to_html(&nl.url).as_bytes())?;
+                        } else {
+                            self.escape_href(url)?;
+                        }
                     }
                     self.output.write_all(b"\" alt=\"")?;
                     return Ok(true);

src/lib.rs

@@ -92,7 +92,7 @@ pub use parser::{
     parse_document, BrokenLinkCallback, BrokenLinkReference, ExtensionOptions,
     ExtensionOptionsBuilder, ListStyleType, Options, ParseOptions, ParseOptionsBuilder, Plugins,
     PluginsBuilder, RenderOptions, RenderOptionsBuilder, RenderPlugins, RenderPluginsBuilder,
-    ResolvedReference,
+    ResolvedReference, URLRewriter,
 };
 pub use typed_arena::Arena;
 pub use xml::format_document as format_xml;

src/parser/mod.rs

@@ -24,6 +24,7 @@ use std::cmp::min;
 use std::collections::HashMap;
 use std::fmt::{self, Debug, Formatter};
 use std::mem;
+use std::panic::RefUnwindSafe;
 use std::str;
 use std::sync::{Arc, Mutex};
 use typed_arena::Arena;
@@ -152,6 +153,18 @@ pub struct Options<'c> {
     pub render: RenderOptions,
 }
 
+/// Trait for link and image URL rewrite extensions.
+pub trait URLRewriter: RefUnwindSafe + Sync {
+    /// Converts the given URL from Markdown to its representation when output as HTML.
+    fn to_html(&self, url: &str) -> String;
+}
+
+impl Debug for dyn URLRewriter {
+    fn fmt(&self, formatter: &mut Formatter<'_>) -> Result<(), std::fmt::Error> {
+        formatter.write_str("<dyn URLRewriter>")
+    }
+}
+
 #[non_exhaustive]
 #[derive(Default, Debug, Clone, Builder)]
 #[cfg_attr(feature = "arbitrary", derive(arbitrary::Arbitrary))]
@@ -521,6 +534,50 @@ pub struct ExtensionOptions {
     /// ```
     #[builder(default)]
     pub greentext: bool,
+
+    /// Wraps embedded image URLs using a custom trait object.
+    ///
+    /// ```
+    /// # use std::sync::Arc;
+    /// # use comrak::{markdown_to_html, ComrakOptions, URLRewriter};
+    /// let mut options = ComrakOptions::default();
+    ///
+    /// struct Rewriter {}
+    /// impl URLRewriter for Rewriter {
+    ///     fn to_html(&self, url: &str) -> String {
+    ///         format!("https://safe.example.com?url={}", url)
+    ///     }
+    /// }
+    ///
+    /// options.extension.image_url_rewriter = Some(Arc::new(Rewriter{}));
+    ///
+    /// assert_eq!(markdown_to_html("![](http://unsafe.example.com/bad.png)", &options),
+    ///            "<p><img src=\"https://safe.example.com?url=http://unsafe.example.com/bad.png\" alt=\"\" /></p>\n");
+    /// ```
+    #[cfg_attr(feature = "arbitrary", arbitrary(value = None))]
+    pub image_url_rewriter: Option<Arc<dyn URLRewriter>>,
+
+    /// Wraps link URLs using a custom trait object.
+    ///
+    /// ```
+    /// # use std::sync::Arc;
+    /// # use comrak::{markdown_to_html, ComrakOptions, URLRewriter};
+    /// let mut options = ComrakOptions::default();
+    ///
+    /// struct Rewriter {}
+    /// impl URLRewriter for Rewriter {
+    ///     fn to_html(&self, url: &str) -> String {
+    ///         format!("https://safe.example.com/norefer?url={}", url)
+    ///     }
+    /// }
+    ///
+    /// options.extension.link_url_rewriter = Some(Arc::new(Rewriter{}));
+    ///
+    /// assert_eq!(markdown_to_html("[my link](http://unsafe.example.com/bad)", &options),
+    ///            "<p><a href=\"https://safe.example.com/norefer?url=http://unsafe.example.com/bad\">my link</a></p>\n");
+    /// ```
+    #[cfg_attr(feature = "arbitrary", arbitrary(value = None))]
+    pub link_url_rewriter: Option<Arc<dyn URLRewriter>>,
 }
 
 #[non_exhaustive]

src/tests.rs

@@ -21,6 +21,7 @@ mod options;
 mod pathological;
 mod plugins;
 mod regressions;
+mod rewriter;
 mod shortcodes;
 mod spoiler;
 mod strikethrough;

src/tests/rewriter.rs

@@ -0,0 +1,36 @@
+use std::sync::Arc;
+
+use super::*;
+
+const IMAGE_PROXY: &'static str = "https://safe.example.com?url=";
+const LINK_NOREFER: &'static str = "https://safe.example.com/norefer?url=";
+
+struct Rewriter {
+    prefix: &'static str,
+}
+
+impl URLRewriter for Rewriter {
+    fn to_html(&self, url: &str) -> String {
+        format!("{}{}", self.prefix, url)
+    }
+}
+
+#[test]
+fn image_url_rewriter() {
+    html_opts_i(
+        "![](http://unsafe.example.com/bad.png)",
+        "<p><img src=\"https://safe.example.com?url=http://unsafe.example.com/bad.png\" alt=\"\" /></p>\n",
+        true,
+        |opts| opts.extension.image_url_rewriter = Some(Arc::new(Rewriter{prefix: IMAGE_PROXY}))
+    );
+}
+
+#[test]
+fn link_url_rewriter() {
+    html_opts_i(
+        "[my link](http://unsafe.example.com/bad)",
+        "<p><a href=\"https://safe.example.com/norefer?url=http://unsafe.example.com/bad\">my link</a></p>\n",
+        true,
+        |opts| opts.extension.link_url_rewriter = Some(Arc::new(Rewriter{prefix: LINK_NOREFER}))
+    );
+}