ui improvements

crates/blend-feed/Cargo.toml

@@ -19,4 +19,5 @@ reqwest = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
 thiserror = { workspace = true }
+tracing = { workspace = true }
 url = { workspace = true }

crates/blend-feed/src/extract/mod.rs

@@ -5,33 +5,3 @@ mod text;
 pub use html::*;
 pub use stylistic::*;
 pub use text::*;
-
-/// Test if given input is likely HTML, based on the difference between the parsed style/inline tags vs. the sanitized HTML.
-pub fn likely_html(src: &str) -> bool {
-    let stylistic = extract_stylistic_html(src, "");
-    let html = extract_html(src, "");
-
-    stylistic != html
-}
-
-#[cfg(test)]
-mod test {
-    use super::*;
-
-    #[test]
-    fn returns_true_for_html() {
-        assert_eq!(likely_html(r#"<p>Some HTML</p>"#), true);
-        assert_eq!(likely_html(r#"<div>Some HTML</div>"#), true);
-        assert_eq!(likely_html(r#"<div>Some <em>HTML</em></div>"#), true);
-    }
-
-    #[test]
-    fn returns_false_for_non_html() {
-        assert_eq!(likely_html(r#"Some HTML"#), false);
-        assert_eq!(likely_html(r#"Some <em>HTML</em>"#), false);
-        assert_eq!(
-            likely_html(r#"<strike><strong>Some</strong> <em>HTML</em></strike>"#),
-            false
-        );
-    }
-}

crates/blend-feed/src/lib.rs

@@ -7,7 +7,7 @@ mod model;
 mod parse;
 mod readability;
 mod scrape;
-mod util;
+pub(crate) mod util;
 
 pub use error::FeedError as Error;
 pub use extract::*;

crates/blend-feed/src/parse/entry.rs

@@ -1,5 +1,5 @@
 use super::{get_feed, parse_url};
-use crate::{error::FeedResult, extract::*, ParsedEntry};
+use crate::{error::FeedResult, extract::*, util::*, ParsedEntry};
 
 /// Number of characters after which we can just assume the summary is the content, without needing to guess that it's HTML
 const MAX_SUMMARY_LENGTH: usize = 1000;
@@ -43,9 +43,9 @@ pub async fn parse_entries(url: &str) -> FeedResult<Vec<ParsedEntry>> {
 
             // Some feeds may return article content as the summary/teaser, so if it's likely HTML, let's swap these fields
             if content_html.is_none()
-                && summary_html
-                    .as_ref()
-                    .is_some_and(|value| value.len() >= MAX_SUMMARY_LENGTH || likely_html(value))
+                && entry.summary.as_ref().is_some_and(|text| {
+                    text.content.len() >= MAX_SUMMARY_LENGTH || likely_html(&text.content)
+                })
             {
                 content_html = entry.summary.map(|text| extract_html(&text.content, base_url));
                 summary_html = None;

crates/blend-feed/src/util/likely_html.rs

@@ -0,0 +1,48 @@
+use crate::{extract_html, extract_stylistic_html};
+
+/// Test if given input is likely HTML, based on the difference between the parsed style/inline tags vs. the sanitized HTML.
+pub fn likely_html(src: &str) -> bool {
+    let stylistic = extract_stylistic_html(src, "");
+    let html = extract_html(src, "");
+
+    stylistic != html
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn returns_true_for_html() {
+        assert_eq!(likely_html(r#"<p>Some HTML</p>"#), true);
+        assert_eq!(likely_html(r#"<div>Some HTML</div>"#), true);
+        assert_eq!(likely_html(r#"<div>Some <em>HTML</em></div>"#), true);
+    }
+
+    #[test]
+    fn returns_false_for_non_html() {
+        assert_eq!(likely_html(r#"Some HTML"#), false);
+        assert_eq!(likely_html(r#"Some <em>HTML</em>"#), false);
+        assert_eq!(
+            likely_html(r#"<strike><strong>Some</strong> <em>HTML</em></strike>"#),
+            false
+        );
+    }
+
+    #[test]
+    fn it_works_on_real_example() {
+        let src = r#"
+            <p class="feature-image">
+                <a href="https://example.com">
+                    <img width="1600" height="840" src="https://example.com/image.jpg" class="attachment-card-large size-card-large wp-post-image" alt="Some image." decoding="async" />
+                </a>
+            </p>
+            <p>Some long description.</p>
+            <p class="read-more">
+                [<a href="https://example.com">Read More</a>]
+            </p>
+        "#;
+
+        assert_eq!(likely_html(src), true);
+    }
+}

crates/blend-feed/src/util/mod.rs

@@ -0,0 +1,5 @@
+mod likely_html;
+mod request;
+
+pub use likely_html::*;
+pub use request::*;

package.json

@@ -15,15 +15,15 @@
     "typeshare:generate": "typeshare ./ --lang typescript --output-file ./ui/src/types/bindings.ts && biome check --apply ./ui/src/types/bindings.ts"
   },
   "dependencies": {
-    "@kobalte/core": "^0.13.1",
+    "@kobalte/core": "^0.13.2",
     "@kobalte/tailwindcss": "^0.9.0",
     "@solid-primitives/active-element": "^2.0.20",
     "@solid-primitives/bounds": "^0.0.121",
     "@solid-primitives/keyboard": "^1.2.8",
     "@solid-primitives/resize-observer": "^2.0.25",
     "@solid-primitives/scheduled": "^1.4.3",
     "@solid-primitives/scroll": "^2.0.23",
-    "@solidjs/router": "^0.13.3",
+    "@solidjs/router": "^0.13.4",
     "@tanstack/solid-query": "^5.40.0",
     "@tanstack/solid-query-devtools": "^5.40.0",
     "class-variance-authority": "^0.7.0",
@@ -34,7 +34,7 @@
   },
   "devDependencies": {
     "@biomejs/biome": "1.7.3",
-    "@swc/core": "^1.5.7",
+    "@swc/core": "^1.5.24",
     "@swc/jest": "^0.2.36",
     "@tailwindcss/forms": "^0.5.7",
     "@tailwindcss/typography": "^0.5.13",
@@ -46,7 +46,7 @@
     "solid-devtools": "^0.30.1",
     "tailwindcss": "^3.4.3",
     "typescript": "^5.4.5",
-    "vite": "^5.2.11",
+    "vite": "^5.2.12",
     "vite-plugin-solid": "^2.10.2",
     "vite-tsconfig-paths": "^4.3.2"
   },