From 9842c8c99c98102a4ec018eaa6403b17c8c34cf5 Mon Sep 17 00:00:00 2001
From: Bill Thiede <git@xinu.tv>
Date: Sat, 25 Jan 2025 16:09:05 -0800
Subject: [PATCH] server: add option to inline CSS before slurping contents

---
 server/src/lib.rs        | 75 +++++++++++++++++++++++++++++++++++++++-
 server/src/newsreader.rs | 16 ++++++---
 web/index.html           |  2 +-
 3 files changed, 87 insertions(+), 6 deletions(-)

diff --git a/server/src/lib.rs b/server/src/lib.rs
index 69ab1a6..4e0bf30 100644
--- a/server/src/lib.rs
+++ b/server/src/lib.rs
@@ -110,6 +110,48 @@ impl Transformer for StripHtml {
     }
 }
 
+struct InlineRemoteStyle<'a> {
+    base_url: &'a Option<Url>,
+}
+
+#[async_trait]
+impl<'a> Transformer for InlineRemoteStyle<'a> {
+    async fn transform(&self, _: &Option<Url>, html: &str) -> Result<String, TransformError> {
+        let css = concat!(
+            "/* chrome-default.css */\n",
+            include_str!("chrome-default.css"),
+            "\n/* mvp.css */\n",
+            include_str!("mvp.css"),
+            "\n/* Xinu Specific overrides */\n",
+            include_str!("custom.css"),
+        );
+        let inline_opts = InlineOptions {
+            //inline_style_tags: true,
+            //keep_style_tags: false,
+            //keep_link_tags: true,
+            base_url: self.base_url.clone(),
+            //load_remote_stylesheets: true,
+            //preallocate_node_capacity: 32,
+            ..InlineOptions::default()
+        };
+
+        //info!("HTML:\n{html}");
+        info!("base_url: {:#?}", self.base_url);
+        Ok(
+            match CSSInliner::options()
+                .base_url(self.base_url.clone())
+                .build()
+                .inline(&html)
+            {
+                Ok(inlined_html) => inlined_html,
+                Err(err) => {
+                    error!("failed to inline remote CSS: {err}");
+                    html.to_string()
+                }
+            },
+        )
+    }
+}
 struct InlineStyle;
 
 #[async_trait]
@@ -229,6 +271,7 @@ impl Transformer for AddOutlink {
 
 struct SlurpContents {
     cacher: Arc<Mutex<FilesystemCacher>>,
+    inline_css: bool,
     site_selectors: HashMap<String, Vec<Selector>>,
 }
 
@@ -267,6 +310,36 @@ impl Transformer for SlurpContents {
             cacher.set(link.as_str(), body.as_bytes());
             body
         };
+        let body = Arc::new(body);
+        let base_url = Some(link.clone());
+        let body = if self.inline_css {
+            let inner_body = Arc::clone(&body);
+            let res = tokio::task::spawn_blocking(move || {
+                let res = CSSInliner::options()
+                    .base_url(base_url)
+                    .build()
+                    .inline(&inner_body);
+
+                match res {
+                    Ok(inlined_html) => inlined_html,
+                    Err(err) => {
+                        error!("failed to inline remote CSS: {err}");
+                        Arc::into_inner(inner_body).expect("failed to take body out of Arc")
+                    }
+                }
+            })
+            .await;
+            match res {
+                Ok(inlined_html) => inlined_html,
+                Err(err) => {
+                    error!("failed to spawn inline remote CSS: {err}");
+                    Arc::into_inner(body).expect("failed to take body out of Arc")
+                }
+            }
+        } else {
+            Arc::into_inner(body).expect("failed to take body out of Arc")
+        };
+
         let doc = Html::parse_document(&body);
 
         let mut results = Vec::new();
@@ -277,7 +350,7 @@ impl Transformer for SlurpContents {
                 //warn!("couldn't find '{:?}' in {}", selector, link);
             }
         }
-        Ok(results.join("<hr>"))
+        Ok(results.join("<br>"))
     }
 }
 
diff --git a/server/src/newsreader.rs b/server/src/newsreader.rs
index c3024d5..586aca8 100644
--- a/server/src/newsreader.rs
+++ b/server/src/newsreader.rs
@@ -15,8 +15,8 @@ use crate::{
     config::Config,
     error::ServerError,
     graphql::{Corpus, NewsPost, Tag, Thread, ThreadSummary},
-    thread_summary_from_row, AddOutlink, EscapeHtml, FrameImages, InlineStyle, Query, SanitizeHtml,
-    SlurpContents, ThreadSummaryRecord, Transformer, NEWSREADER_TAG_PREFIX,
+    thread_summary_from_row, AddOutlink, EscapeHtml, FrameImages, InlineRemoteStyle, Query,
+    SanitizeHtml, SlurpContents, ThreadSummaryRecord, Transformer, NEWSREADER_TAG_PREFIX,
     NEWSREADER_THREAD_PREFIX,
 };
 
@@ -196,6 +196,8 @@ pub async fn thread(
     let body_tranformers: Vec<Box<dyn Transformer>> = vec![
         Box::new(SlurpContents {
             cacher,
+            // TODO: make this true when bulma is finally removed
+            inline_css: false,
             site_selectors: hashmap![
                 "atmeta.com".to_string() => vec![
                     Selector::parse("div.entry-content").unwrap(),
@@ -223,6 +225,9 @@ pub async fn thread(
                 "ingowald.blog".to_string() => vec![
                     Selector::parse("article").unwrap(),
                 ],
+                "jvns.ca".to_string() => vec![
+                    Selector::parse("article").unwrap(),
+                ],
                 "mitchellh.com".to_string() => vec![Selector::parse("div.w-full").unwrap()],
                 "natwelch.com".to_string() => vec![
                     Selector::parse("article div.prose").unwrap(),
@@ -234,6 +239,9 @@ pub async fn thread(
                     Selector::parse("span.story-byline").unwrap(),
                     Selector::parse("div.p").unwrap(),
                 ],
+                "trofi.github.io".to_string() => vec![
+                    Selector::parse("#content").unwrap(),
+                ],
                 "www.redox-os.org".to_string() => vec![
                     Selector::parse("div.content").unwrap(),
                 ],
@@ -245,12 +253,12 @@ pub async fn thread(
         }),
         Box::new(FrameImages),
         Box::new(AddOutlink),
-        Box::new(EscapeHtml),
+        // TODO: causes doubling of images in cloudflare blogs
+        //Box::new(EscapeHtml),
         Box::new(SanitizeHtml {
             cid_prefix: "",
             base_url: &link,
         }),
-        Box::new(InlineStyle),
     ];
     for t in body_tranformers.iter() {
         if t.should_run(&link, &body) {
diff --git a/web/index.html b/web/index.html
index 5772e5e..a90ddb7 100644
--- a/web/index.html
+++ b/web/index.html
@@ -26,4 +26,4 @@
   <section id="app"></section>
 </body>
 
-</html>
+</html>
\ No newline at end of file