generate <meta> og:image in individual thread pages (#21)

src/lib.rs

@@ -66,6 +66,7 @@ pub struct ExtractedPost {
     pub dom: RcDom,
     pub meta: PostMeta,
     pub needs_attachments: BTreeSet<SitePath>,
+    pub og_image: Option<String>,
 }
 
 #[derive(Clone, Debug)]
@@ -74,6 +75,7 @@ pub struct Thread {
     pub posts: Vec<TemplatedPost>,
     pub meta: PostMeta,
     pub needs_attachments: BTreeSet<SitePath>,
+    pub og_image: Option<String>,
 }
 
 #[derive(Clone, Debug)]
@@ -83,6 +85,7 @@ pub struct TemplatedPost {
     pub original_html: String,
     pub safe_html: String,
     pub needs_attachments: BTreeSet<SitePath>,
+    pub og_image: Option<String>,
 }
 
 impl Thread {
@@ -199,21 +202,24 @@ impl TryFrom<TemplatedPost> for Thread {
         // TODO: skip threads with private or logged-in-only authors?
         // TODO: gate sensitive posts behind an interaction?
 
-        // for the thread title, take the last post that is not a transparent share (which MAY have
-        // tags, but SHOULD NOT have a title and MUST NOT have a body), and use its title if any.
-        meta.title = posts
+        // for thread metadata, take the last post that is not a transparent share (which MAY have
+        // tags, but SHOULD NOT have a title and MUST NOT have a body), and use its metadata if any.
+        let last_non_transparent_share_post = posts
             .iter()
             .rev()
-            .find(|post| !post.meta.is_transparent_share)
-            .map(|post| {
-                if let Some(title) = post.meta.title.clone().filter(|t| !t.is_empty()) {
-                    title
-                } else if let Some(author) = post.meta.author.as_ref() {
-                    format!("untitled post by {}", author.display_handle)
-                } else {
-                    "untitled post".to_owned()
-                }
-            });
+            .find(|post| !post.meta.is_transparent_share);
+        meta.title = last_non_transparent_share_post.map(|post| {
+            if let Some(title) = post.meta.title.clone().filter(|t| !t.is_empty()) {
+                title
+            } else if let Some(author) = post.meta.author.as_ref() {
+                format!("untitled post by {}", author.display_handle)
+            } else {
+                "untitled post".to_owned()
+            }
+        });
+        let og_image = last_non_transparent_share_post
+            .and_then(|post| post.og_image.as_deref())
+            .map(|og_image| SETTINGS.base_url_relativise(og_image));
 
         let needs_attachments = posts
             .iter()
@@ -226,6 +232,7 @@ impl TryFrom<TemplatedPost> for Thread {
             posts,
             meta,
             needs_attachments,
+            og_image,
         })
     }
 }
@@ -272,6 +279,7 @@ impl TemplatedPost {
             original_html: unsafe_html.to_owned(),
             safe_html,
             needs_attachments: post.needs_attachments,
+            og_image: post.og_image,
         })
     }
 }

src/meta.rs

@@ -19,6 +19,7 @@ pub fn extract_metadata(unsafe_html: &str) -> eyre::Result<ExtractedPost> {
 
     let mut meta = PostMeta::default();
     let mut needs_attachments = BTreeSet::default();
+    let mut og_image = None;
     let mut author_href = None;
     let mut author_name = None;
     let mut author_display_name = None;
@@ -74,17 +75,25 @@ pub fn extract_metadata(unsafe_html: &str) -> eyre::Result<ExtractedPost> {
                             _ => {}
                         }
                         continue;
-                    } else if let Some(attr_names) = html_attributes_with_urls().get(name) {
-                        for attr in attrs.iter() {
-                            if attr_names.contains(&attr.name) {
-                                if let Ok(url) =
-                                    SitePath::from_rendered_attachment_url(attr.value.to_str())
-                                {
-                                    trace!("found attachment url in rendered post: {url:?}");
-                                    needs_attachments.insert(url);
+                    } else {
+                        if let Some(attr_names) = html_attributes_with_urls().get(name) {
+                            for attr in attrs.iter() {
+                                if attr_names.contains(&attr.name) {
+                                    if let Ok(url) =
+                                        SitePath::from_rendered_attachment_url(attr.value.to_str())
+                                    {
+                                        trace!("found attachment url in rendered post: {url:?}");
+                                        needs_attachments.insert(url);
+                                    }
                                 }
                             }
                         }
+                        // use the first <img src>, if any, as the <meta> og:image.
+                        if og_image.is_none() && name == &QualName::html("img") {
+                            if let Some(src) = attrs.attr_str("src")?.map(|t| t.to_owned()) {
+                                og_image = Some(src);
+                            }
+                        }
                     }
                 }
                 _ => {}
@@ -111,6 +120,7 @@ pub fn extract_metadata(unsafe_html: &str) -> eyre::Result<ExtractedPost> {
         dom,
         meta,
         needs_attachments,
+        og_image,
     })
 }
 

templates/thread-or-post-meta.html

@@ -1 +1,2 @@
 <meta property="og:title" content="{% if let Some(title) = thread.meta.title %}{{ title }}{% endif %}">
+{%~ if let Some(og_image) = thread.og_image ~%}<meta property="og:image" content="{{ og_image }}">{%~ endif ~%}