Show metadata size

Fixes #2637

Author: hyanwong

python/CHANGELOG.rst

@@ -8,6 +8,11 @@
   ID is e.g. a population name, rather than silently returning no samples.
   (:user:`hyanwong`, :pr:`3344`)
 
+**Features**
+
+- Displaying a summary of the tree sequence now shows the metadata codec and
+  size of the metadata for each table. (:user:`hyanwong`, :pr:`3343`, :issue:`2637`)
+
 --------------------
 [1.0.0] - 2025-11-27
 --------------------

python/tests/test_highlevel.py

@@ -1991,8 +1991,25 @@ def test_html_repr(self, ts):
         assert len(html) > 5000
         assert f"<tr><td>Trees</td><td>{ts.num_trees:,}</td></tr>" in html
         assert f"<tr><td>Time Units</td><td>{ts.time_units}</td></tr>" in html
-        for table in ts.tables.table_name_map:
-            assert f"<td>{table.capitalize()}</td>" in html
+        codecs = collections.defaultdict(int)
+        for table_name, table in ts.tables.table_name_map.items():
+            assert f"<td>{table_name.capitalize()}</td>" in html
+            if hasattr(table, "metadata_schema"):
+                schema = table.metadata_schema.schema
+                codec = schema["codec"] if schema else "raw"
+                codecs[codec] += 1
+        assert "<td>Metadata</td>" in html
+        assert "<th>Metadata</th>" in html
+        assert "<th>Metadata size</th>" in html
+        num_tables_with_metadata = 0
+        for codec, count in codecs.items():
+            assert html.count(f">{codec}</td>") == count
+            num_tables_with_metadata += count
+        # Only one table (provenances) has no metadata
+        assert num_tables_with_metadata == len(ts.tables.table_name_map) - 1
+        # All metadata tables should show the percentage metadata size
+        assert html.count("%)</td>") == num_tables_with_metadata
+
         if ts.num_provenances > 0:
             assert (
                 f"<td>{json.loads(ts.provenance(0).record)['software']['name']}</td>"

python/tskit/trees.py

@@ -4476,19 +4476,16 @@ def __str__(self):
             ["Sample Nodes", util.format_number(self.num_samples, sep=",")],
             ["Total Size", util.naturalsize(self.nbytes)],
         ]
-        header = ["Table", "Rows", "Size", "Has Metadata"]
+        header = ["Table", "Rows", "Size", "Metadata", "Metadata size"]
         table_rows = []
         for name, table in self.tables.table_name_map.items():
             table_rows.append(
                 [
                     name.capitalize(),
                     f"{util.format_number(table.num_rows, sep=',')}",
                     util.naturalsize(table.nbytes),
-                    (
-                        "Yes"
-                        if hasattr(table, "metadata") and len(table.metadata) > 0
-                        else "No"
-                    ),
+                    util.metadata_codec(table),
+                    util.metadata_size(table),
                 ]
             )
         return util.unicode_table(ts_rows, title="TreeSequence") + util.unicode_table(

python/tskit/util.py

@@ -534,17 +534,29 @@ def html_table(rows, *, header):
     """
 
 
+def metadata_codec(table):
+    if hasattr(table, "metadata_schema"):
+        schema = table.metadata_schema.schema
+        return "raw" if schema is None else schema.get("codec", "unknown")
+    return ""
+
+
+def metadata_size(table):
+    if hasattr(table, "metadata"):
+        frac = len(table.metadata) / table.nbytes
+        return f"{naturalsize(len(table.metadata))} ({frac:.0%})"
+    return ""
+
+
 def tree_sequence_html(ts):
     table_rows = "".join(
         f"""
             <tr>
                 <td>{name.capitalize()}</td>
                 <td>{format_number(table.num_rows)}</td>
                 <td>{naturalsize(table.nbytes)}</td>
-                <td style="text-align: center;">
-                    {'✅' if hasattr(table, "metadata") and len(table.metadata) > 0
-                     else ''}
-                </td>
+                <td style="text-align: center;">{metadata_codec(table)}</td>
+                <td>{metadata_size(table)}</td>
             </tr>
         """
         for name, table in ts.tables.table_name_map.items()
@@ -637,7 +649,8 @@ def tree_sequence_html(ts):
                                 <th style="line-height:21px;">Table</th>
                                 <th>Rows</th>
                                 <th>Size</th>
-                                <th>Has Metadata</th>
+                                <th>Metadata</th>
+                                <th>Metadata size</th>
                             </tr>
                         </thead>
                         <tbody>