Bump ord-schema to 0.7 and sort datasets by submission date (#200)

ord-schema 0.7.0 splits its heavy optional dependencies into extras and
adds a denormalized `dataset.submitted_at` column.

- Depend on `ord-schema[orm]>=0.7,<0.8` so only the ORM extra (sqlalchemy,
  psycopg, inflection) is pulled in. This drops huggingface-hub and its
  transitive deps (hf-xet, fsspec, rich, typer, ...) from the install —
  9 fewer packages (92 -> 83) and a smaller image. pygithub, still needed
  by the editor, remains a direct dependency.
- Surface `submitted_at` (a nullable date) in `DatasetInfo` and order the
  `/datasets` response by it, newest first, with undated datasets last.
- Add a "Submitted" column to the browse page so the ordering is visible.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

Author: skearnes

app/src/views/browse/MainBrowse.scss

@@ -21,7 +21,7 @@
   padding: 1rem 0;
 
   .table-container {
-    grid-template-columns: 1fr 1fr 1fr auto;
+    grid-template-columns: 1fr 1fr 1fr auto auto;
   }
 
   .loading {

app/src/views/browse/MainBrowse.tsx

@@ -25,6 +25,7 @@ interface Dataset {
   name: string;
   description?: string;
   num_reactions: number;
+  submitted_at?: string | null;
 }
 
 const MainBrowse: React.FC = () => {
@@ -81,6 +82,7 @@ const MainBrowse: React.FC = () => {
             <div className="column label">Name</div>
             <div className="column label">Description</div>
             <div className="column label">Size</div>
+            <div className="column label">Submitted</div>
             {entities.map(row => (
               <React.Fragment key={row.dataset_id}>
                 <div className="column">
@@ -89,6 +91,7 @@ const MainBrowse: React.FC = () => {
                 <div className="column">{row.name}</div>
                 <div className="column">{truncateDescription(row.description)}</div>
                 <div className="column">{row.num_reactions}</div>
+                <div className="column">{row.submitted_at ?? '—'}</div>
               </React.Fragment>
             ))}
           </div>

ord_interface/api/search.py

@@ -16,6 +16,7 @@
 
 from __future__ import annotations
 
+import datetime
 import gzip
 import json
 import os
@@ -209,14 +210,16 @@ class DatasetInfo(BaseModel):
     name: str
     description: str | None
     num_reactions: int
+    submitted_at: datetime.date | None
 
 
 @router.get("/datasets")
 async def get_datasets() -> list[DatasetInfo]:
-    """Returns info about the current datasets."""
+    """Returns info about the current datasets, most recently submitted first."""
     async with get_cursor() as cursor:
         await cursor.execute(
-            "SELECT dataset_id, name, description, num_reactions FROM dataset"
+            "SELECT dataset_id, name, description, num_reactions, submitted_at FROM dataset "
+            "ORDER BY submitted_at DESC NULLS LAST, dataset_id"
         )
         return [DatasetInfo(**row) async for row in cursor]
 
@@ -226,7 +229,7 @@ async def get_dataset(dataset_id: str) -> DatasetInfo:
     """Returns info about a single dataset."""
     async with get_cursor() as cursor:
         await cursor.execute(
-            "SELECT dataset_id, name, description, num_reactions FROM dataset WHERE dataset_id = %s",
+            "SELECT dataset_id, name, description, num_reactions, submitted_at FROM dataset WHERE dataset_id = %s",
             (dataset_id,),
         )
         row = await cursor.fetchone()

ord_interface/api/search_test.py

@@ -91,6 +91,13 @@ def test_get_datasets(test_client):
     response.raise_for_status()
     dataset_info = response.json()
     assert len(dataset_info) == 3
+    assert all("submitted_at" in row for row in dataset_info)
+    # Datasets are ordered by submission date, newest first, with undated
+    # datasets (submitted_at is None) sorted last.
+    submitted = [row["submitted_at"] for row in dataset_info]
+    dated = [value for value in submitted if value is not None]
+    assert dated == sorted(dated, reverse=True)
+    assert submitted == dated + [None] * (len(submitted) - len(dated))
 
 
 def test_get_dataset(test_client):

pyproject.toml

@@ -31,7 +31,7 @@ dependencies = [
     "gunicorn",
     "jinja2>=2.0.0",
     "numpy>=1.26.4",
-    "ord-schema>=0.6.32,<0.7",
+    "ord-schema[orm]>=0.7,<0.8",
     "pandas>=1.0.4",
     "protobuf>=4.22,<6",
     "psycopg[binary,pool]>=3",