Add ADDITIONAL_HF_REPOS config option for Hugging Face repos outside the org

README.md

@@ -82,6 +82,7 @@ Welcome to your new catalog repo! The primary way to personalize this catalog is
   * `API_BASE_URL`: Hugging Face API base URL (default: `"https://huggingface.co/api/"`)
   * `REFRESH_INTERVAL_DAYS`: Number of days to consider an item "new" (default: `30`)
   * `ADDITIONAL_REPOS`: Array of forked or non-org GitHub repositories to include, formatted `<owner>/<repo-name>` (non-forks are included by default). Use `[]` if there are none you wish to include
+  * `ADDITIONAL_HF_REPOS`: Array of Hugging Face repos from outside the org to include. Each entry specifies `repo` (`<owner>/<repo-name>`) and `type` (`datasets`, `models`, or `spaces`). Use `[]` if there are none you wish to include
 
 * **Typography:**
   * `FONT_FAMILY`: Font family for the site (default: `"Inter"`)

main.js

@@ -19,7 +19,7 @@ const configPromise = fetch('config.yaml')
 
 // Module-scope lets — assigned after config loads, used by all functions below
 let CONFIG;
-let ORGANIZATION_NAME, CATALOG_REPO_NAME, API_BASE_URL, REFRESH_INTERVAL_DAYS, ADDITIONAL_REPOS;
+let ORGANIZATION_NAME, CATALOG_REPO_NAME, API_BASE_URL, REFRESH_INTERVAL_DAYS, ADDITIONAL_REPOS, ADDITIONAL_HF_REPOS;
 
 // Build a reverse lookup from TAG_GROUPS (defined in tag-groups.js): raw tag → [canonical tags]
 // A raw tag may appear in multiple groups, so the value is an array.
@@ -296,6 +296,36 @@ const fetchHubItems = async (repoType) => {
         }
         let hfItems = await response.json();
 
+        // Fetch additional HF repos of this type from outside the org
+        const additionalForType = ADDITIONAL_HF_REPOS.filter(entry => entry.type === repoType);
+        if (additionalForType.length) {
+            const existingIds = new Set(hfItems.map(item => item.id));
+            const seenRepos = new Set();
+            const toFetch = additionalForType.filter(entry => {
+                if (existingIds.has(entry.repo) || seenRepos.has(entry.repo)) return false;
+                seenRepos.add(entry.repo);
+                return true;
+            });
+
+            const fetched = await Promise.all(
+                toFetch.map(entry =>
+                    fetch(`${API_BASE_URL}${repoType}/${entry.repo}`)
+                        .then(r => {
+                            if (!r.ok) {
+                                console.warn(`Failed to fetch additional HF repo "${entry.repo}": HTTP ${r.status}`);
+                                return null;
+                            }
+                            return r.json();
+                        })
+                        .catch(err => {
+                            console.warn(`Network error fetching additional HF repo "${entry.repo}":`, err);
+                            return null;
+                        })
+                )
+            );
+            hfItems = [...hfItems, ...fetched.filter(item => item && !existingIds.has(item.id))];
+        }
+
         // Step 2: If we are fetching models, get the full details for each one.
         if (repoType === 'models') {
             const detailPromises = hfItems.map(item =>
@@ -731,7 +761,18 @@ document.addEventListener('DOMContentLoaded', async () => {
         if (!CONFIG.ORGANIZATION_NAME)                 missing.push('ORGANIZATION_NAME');
         if (!CONFIG.API_BASE_URL)                      missing.push('API_BASE_URL');
         if (CONFIG.REFRESH_INTERVAL_DAYS == null)      missing.push('REFRESH_INTERVAL_DAYS');
-        if (!Array.isArray(CONFIG.ADDITIONAL_REPOS))   missing.push('ADDITIONAL_REPOS (must be a list)');
+        if (!Array.isArray(CONFIG.ADDITIONAL_REPOS))    missing.push('ADDITIONAL_REPOS (must be a list)');
+        if (!Array.isArray(CONFIG.ADDITIONAL_HF_REPOS)) {
+            missing.push('ADDITIONAL_HF_REPOS (must be a list)');
+        } else {
+            const validTypes = new Set(['datasets', 'models', 'spaces']);
+            const badEntries = CONFIG.ADDITIONAL_HF_REPOS.filter(
+                e => !e || typeof e.repo !== 'string' || !e.repo.trim() || !validTypes.has(e.type)
+            );
+            if (badEntries.length) missing.push(
+                `ADDITIONAL_HF_REPOS entries must each have a non-empty "repo" string and "type" in {datasets, models, spaces}; bad entries: ${badEntries.map(e => JSON.stringify(e)).join(', ')}`
+            );
+        }
         if (!CONFIG.COLORS || typeof CONFIG.COLORS !== 'object') {
             missing.push('COLORS (must be an object with primary, secondary, accent, accentDark, tag)');
         } else {
@@ -754,7 +795,7 @@ document.addEventListener('DOMContentLoaded', async () => {
             LOGO_URL: '', FAVICON_URL: '',
             COLORS: { primary: '#92991c', secondary: '#5d8095', accent: '#0097b2', accentDark: '#4fd1eb', tag: '#9bcb5e' },
             API_BASE_URL: 'https://huggingface.co/api/', REFRESH_INTERVAL_DAYS: 30,
-            ADDITIONAL_REPOS: [], FONT_FAMILY: 'Inter'
+            ADDITIONAL_REPOS: [], ADDITIONAL_HF_REPOS: [], FONT_FAMILY: 'Inter'
         };
     }
 
@@ -764,6 +805,7 @@ document.addEventListener('DOMContentLoaded', async () => {
     API_BASE_URL          = CONFIG.API_BASE_URL;
     REFRESH_INTERVAL_DAYS = CONFIG.REFRESH_INTERVAL_DAYS;
     ADDITIONAL_REPOS      = CONFIG.ADDITIONAL_REPOS;
+    ADDITIONAL_HF_REPOS   = CONFIG.ADDITIONAL_HF_REPOS;
 
     // Apply CSS custom properties and document metadata
     document.title = CONFIG.CATALOG_TITLE || 'Catalog';

public/config.yaml

@@ -34,5 +34,16 @@ ADDITIONAL_REPOS:
   - "jennamk14/autonomous_drone_simulator"
   - "ixlab/vidformer"
 
+# Array of Hugging Face repos from outside the org to include.
+# Each entry must specify "repo" (owner/name) and "type" (datasets, models, or spaces).
+# ADDITIONAL_HF_REPOS:
+#   - repo: "user/dataset-name"
+#     type: "datasets"
+#   - repo: "user/model-name"
+#     type: "models"
+#   - repo: "user/space-name"
+#     type: "spaces"
+ADDITIONAL_HF_REPOS: []
+
 # Typography
 FONT_FAMILY: Inter    # Font family for the site

scripts/export-tags.js

@@ -45,6 +45,21 @@ if (!Array.isArray(rawConfig.ADDITIONAL_REPOS)) {
 
 const CONFIG = rawConfig;
 const { ORGANIZATION_NAME, API_BASE_URL, ADDITIONAL_REPOS } = CONFIG;
+if (!Array.isArray(CONFIG.ADDITIONAL_HF_REPOS)) {
+    throw new Error(
+        `Invalid config at ${configPath}: ADDITIONAL_HF_REPOS must be an array.`
+    );
+}
+const validHFTypes = new Set(['datasets', 'models', 'spaces']);
+const badHFEntries = CONFIG.ADDITIONAL_HF_REPOS.filter(
+    e => !e || typeof e.repo !== 'string' || !e.repo.trim() || !validHFTypes.has(e.type)
+);
+if (badHFEntries.length) {
+    throw new Error(
+        `Invalid config at ${configPath}: ADDITIONAL_HF_REPOS entries must each have a non-empty "repo" string and "type" in {datasets, models, spaces}; bad entries: ${badHFEntries.map(e => JSON.stringify(e)).join(', ')}`
+    );
+}
+const ADDITIONAL_HF_REPOS = CONFIG.ADDITIONAL_HF_REPOS;
 
 // ---------------------------------------------------------------------------
 // Fetch helpers
@@ -104,6 +119,26 @@ const collectHFTags = async (repoType) => {
     console.log(`Fetching HF ${repoType}...`);
     let items = (await get(`${API_BASE_URL}${repoType}?author=${ORGANIZATION_NAME}&full=true`)).json;
 
+    // Fetch additional HF repos of this type
+    const additionalForType = ADDITIONAL_HF_REPOS.filter(entry => entry.type === repoType);
+    if (additionalForType.length) {
+        const existingIds = new Set(items.map(item => item.id));
+        const seenRepos = new Set();
+        const toFetch = additionalForType.filter(entry => {
+            if (existingIds.has(entry.repo) || seenRepos.has(entry.repo)) return false;
+            seenRepos.add(entry.repo);
+            return true;
+        });
+        const fetched = await Promise.all(
+            toFetch.map(entry =>
+                get(`${API_BASE_URL}${repoType}/${entry.repo}`)
+                    .then(({ json }) => json)
+                    .catch(() => null)
+            )
+        );
+        items = [...items, ...fetched.filter(item => item && !existingIds.has(item.id))];
+    }
+
     if (repoType === 'models') {
         const details = await Promise.all(
             items.map(item =>