Fix canonical media links in preview and generation

crates/bds-core/src/engine/preview.rs

@@ -1218,6 +1218,105 @@ mod tests {
         assert!(pagefind_body.contains("window.pagefind"));
     }
 
+    #[test]
+    fn preview_rewrites_media_aliases_and_serves_the_canonical_file() {
+        let _guard = preview_port_guard().lock().unwrap();
+        let (dir, db) = setup_preview_fixture();
+        std::fs::create_dir_all(dir.path().join("media/2022/11")).unwrap();
+        std::fs::write(
+            dir.path().join("media/2022/11/canonical-id.jpg"),
+            b"canonical image",
+        )
+        .unwrap();
+        queries::media::insert_media(
+            db.conn(),
+            &Media {
+                id: "canonical-id".into(),
+                project_id: "project-1".into(),
+                filename: "canonical-id.jpg".into(),
+                original_name: "20221111_0177.jpg".into(),
+                mime_type: "image/jpeg".into(),
+                size: 15,
+                width: Some(100),
+                height: Some(80),
+                title: None,
+                alt: None,
+                caption: None,
+                author: None,
+                language: None,
+                file_path: "media/2022/11/canonical-id.jpg".into(),
+                sidecar_path: "media/2022/11/canonical-id.jpg.meta".into(),
+                checksum: None,
+                tags: Vec::new(),
+                created_at: 1_668_124_800_000,
+                updated_at: 1_668_124_800_000,
+            },
+        )
+        .unwrap();
+        let markdown = "[Open image](/media/2022/11/20221111_0177.jpg?download=1#preview)\n\n![Embedded image](bds-media://canonical-id)";
+        let mut post = make_draft_post();
+        post.content = Some(markdown.into());
+        queries::post::insert_post(db.conn(), &post).unwrap();
+
+        let server = start_preview_server(
+            dir.path().join("bds.db"),
+            dir.path().to_path_buf(),
+            "project-1".into(),
+        )
+        .unwrap();
+        let client = reqwest::blocking::Client::new();
+        let preview_html = client
+            .get(format!(
+                "http://{PREVIEW_HOST}:{PREVIEW_PORT}/__draft/post-1"
+            ))
+            .send()
+            .unwrap()
+            .text()
+            .unwrap();
+        let image_response = client
+            .get(format!(
+                "http://{PREVIEW_HOST}:{PREVIEW_PORT}/media/2022/11/canonical-id.jpg?download=1"
+            ))
+            .send()
+            .unwrap();
+        let image_status = image_response.status();
+        let image_content_type = image_response
+            .headers()
+            .get(header::CONTENT_TYPE)
+            .unwrap()
+            .to_str()
+            .unwrap()
+            .to_string();
+        let image_body = image_response.bytes().unwrap();
+        server.stop().unwrap();
+
+        let mut published_post = post;
+        published_post.status = PostStatus::Published;
+        let generated = crate::render::build_site_render_artifacts(
+            db.conn(),
+            dir.path(),
+            "project-1",
+            &make_metadata(),
+            &[(published_post, markdown.into())],
+        )
+        .unwrap();
+        let generated_html = &generated
+            .pages
+            .iter()
+            .find(|page| page.url_path == "/2024/03/09/hello")
+            .unwrap()
+            .html;
+        let expected = "href=\"/media/2022/11/canonical-id.jpg?download=1#preview\"";
+
+        assert!(preview_html.contains(expected));
+        assert!(generated_html.contains(expected));
+        assert!(preview_html.contains("src=\"/media/2022/11/canonical-id.jpg\""));
+        assert!(generated_html.contains("src=\"/media/2022/11/canonical-id.jpg\""));
+        assert_eq!(image_status, StatusCode::OK);
+        assert_eq!(image_content_type, "image/jpeg");
+        assert_eq!(image_body.as_ref(), b"canonical image");
+    }
+
     #[test]
     fn preview_server_serves_style_preview() {
         let _guard = preview_port_guard().lock().unwrap();

crates/bds-core/src/render/site.rs

@@ -22,7 +22,7 @@ use crate::render::{
 };
 use crate::scripting::{CoreHost, HostApi, UnavailableHost};
 use crate::util::frontmatter::{read_script_file, read_template_file, read_translation_file};
-use crate::util::slugify;
+use crate::util::{slugify, year_month_from_unix_ms};
 
 const STARTER_SINGLE_POST_TEMPLATE: &str =
     include_str!("../../../../assets/starter-templates/single-post.liquid");
@@ -187,6 +187,7 @@ fn build_site_render_artifacts_with_mode(
         .map(|media| (media.id.clone(), media))
         .collect::<HashMap<_, _>>();
     let project_media = media_items.iter().map(media_context).collect::<Vec<_>>();
+    let canonical_media_map = canonical_media_paths(&media_items);
     let project_tags = build_published_tag_counts(published_posts, &tags);
 
     let mut artifacts = SiteRenderArtifacts::default();
@@ -238,6 +239,7 @@ fn build_site_render_artifacts_with_mode(
                     &category_settings,
                     &menu_items,
                     &post_data_json_by_id,
+                    &canonical_media_map,
                     &project_media,
                     &project_tags,
                     &bundle,
@@ -331,6 +333,7 @@ fn build_site_render_artifacts_with_mode(
                     &canonical_map,
                     &menu_items,
                     &post_data_json_by_id,
+                    &canonical_media_map,
                     &project_media,
                     &project_tags,
                     &bundle,
@@ -805,6 +808,7 @@ fn render_list_route(
     category_settings: &HashMap<String, CategorySettings>,
     menu_items: &[Value],
     post_data_json_by_id: &HashMap<String, Value>,
+    canonical_media_path_by_source_path: &HashMap<String, String>,
     project_media: &[Value],
     project_tags: &[Value],
     bundle: &TemplateBundle,
@@ -849,7 +853,7 @@ fn render_list_route(
         "total_items": route.total_items,
         "items_per_page": route.items_per_page,
         "canonical_post_path_by_slug": canonical_map,
-        "canonical_media_path_by_source_path": canonical_media_paths(posts),
+        "canonical_media_path_by_source_path": canonical_media_path_by_source_path,
         "post_data_json_by_id": post_data_json_by_id,
         "project": { "media": project_media },
         "Tags": project_tags,
@@ -883,6 +887,7 @@ fn render_post_route(
     canonical_post_path_by_slug: &HashMap<String, String>,
     menu_items: &[Value],
     post_data_json_by_id: &HashMap<String, Value>,
+    canonical_media_path_by_source_path: &HashMap<String, String>,
     project_media: &[Value],
     project_tags: &[Value],
     bundle: &TemplateBundle,
@@ -959,7 +964,7 @@ fn render_post_route(
         "tag_color_by_name": taxonomy_counts.2,
         "backlinks": backlinks,
         "canonical_post_path_by_slug": canonical_post_path_by_slug,
-        "canonical_media_path_by_source_path": canonical_media_paths(all_posts),
+        "canonical_media_path_by_source_path": canonical_media_path_by_source_path,
         "post_data_json_by_id": post_data_json_by_id,
         "project": { "media": project_media },
         "Tags": project_tags,
@@ -1229,31 +1234,22 @@ fn canonical_post_path_by_slug(
         .collect()
 }
 
-fn canonical_media_paths(posts: &[RenderPostRecord]) -> HashMap<String, String> {
+fn canonical_media_paths(media_items: &[Media]) -> HashMap<String, String> {
     let mut map = HashMap::new();
-    for record in posts {
+    for media in media_items {
-        for media_reference in extract_media_refs(&record.body_markdown) {
+        let target = canonical_media_path(media);
-            map.entry(media_reference.clone())
+        let (year, month) = year_month_from_unix_ms(media.created_at);
-                .or_insert_with(|| media_reference.clone());
+        for source in [
+            format!("media/{year}/{month}/{}", media.original_name).to_lowercase(),
+            format!("bds-media://{}", media.id),
+            media.file_path.trim_start_matches('/').to_lowercase(),
+        ] {
+            map.entry(source).or_insert_with(|| target.clone());
         }
     }
     map
 }
 
-fn extract_media_refs(markdown: &str) -> Vec<String> {
-    let mut refs = Vec::new();
-    let mut remainder = markdown;
-    while let Some(index) = remainder.find("bds-media://") {
-        let rest = &remainder[index..];
-        let end = rest
-            .find(|ch: char| ch == ')' || ch == ']' || ch.is_whitespace())
-            .unwrap_or(rest.len());
-        refs.push(rest[..end].to_string());
-        remainder = &rest[end..];
-    }
-    refs
-}
-
 fn build_linked_media_by_post_id(
     conn: &Connection,
     posts: &[RenderPostRecord],