FIX better display of object columns with multi-line repr in tablerep…

…ort (#1196)

CHANGES.rst

@@ -50,6 +50,10 @@ Bug fixes
   :user:`Jérôme Dockès <jeromedockes>` and the matplotlib issue can be tracked
   [here](https://github.com/matplotlib/matplotlib/issues/25041).
 
+* The labels on bar plots in the ``TableReport`` for columns of object dtypes
+  that have a repr spanning multiple lines could be unreadable. This has been
+  fixed in :pr:`1196` by :user:`Jérôme Dockès <jeromedockes>`.
+
 * Improve the performance of :func:`deduplicate` by removing some unnecessary
   computations. :pr:`1193` by :user:`Jérôme Dockès <jeromedockes>`.
 

skrub/_reporting/_utils.py

@@ -43,8 +43,8 @@ def quantiles(column):
 
 def ellide_string(s, max_len=30):
     """Shorten a string so it can be used as a plot axis title or label."""
-    if not isinstance(s, str):
-        return s
+    s = str(s)
+
     # normalize whitespace
     s = re.sub(r"\s+", " ", s)
     if len(s) <= max_len:

skrub/_reporting/tests/test_utils.py

@@ -11,7 +11,7 @@
 @pytest.mark.parametrize(
     "s_in, s_out",
     [
-        (1, 1),
+        (1, "1"),
         ("aa", "aa"),
         ("a\na", "a a"),
         ("a" * 70, "a" * 30 + "…\u200e"),
@@ -55,6 +55,16 @@ def test_ellide_string_empty():
     assert _utils.ellide_string(" a", 1) == "…"
 
 
+def test_ellide_non_string():
+    # non-regression for #1195: objects in columns must be converted to strings
+    # before elliding and plotting
+    class A:
+        def __repr__(self):
+            return "one\ntwo\nthree"
+
+    assert _utils.ellide_string(A()) == "one two three"
+
+
 @pytest.mark.parametrize(
     "n_in, n_out",
     [