From 0fab233091897458b5832d8364baeb6148019c1a Mon Sep 17 00:00:00 2001
From: Adrian Letchford <me@dradrian.com>
Date: Wed, 15 Oct 2025 16:51:33 +0000
Subject: [PATCH] Print the top and bottom bad rows

---
 src/recx/results.py           | 18 ++++++++++++++++--
 tests/fixtures/frames.py      | 20 ++++++++++++++++++++
 tests/integration/test_rec.py | 14 ++++++++++++++
 3 files changed, 50 insertions(+), 2 deletions(-)

diff --git a/src/recx/results.py b/src/recx/results.py
index dc32340..5d152a9 100644
--- a/src/recx/results.py
+++ b/src/recx/results.py
@@ -7,6 +7,20 @@
 logger = logging.getLogger(__name__)
 
 
+def df2str(
+    df: pd.DataFrame,
+    max_rows: int | None = None,
+) -> str:
+    args = []
+    args += ["display.min_rows", max_rows]
+    args += ["display.max_rows", max_rows]
+    args += ["display.max_columns", None]
+    args += ["display.width", 10000]
+
+    with pd.option_context(*args):
+        return df.__str__()
+
+
 class CheckResult:
     """
     Result of an individual column (or index) check.
@@ -42,7 +56,7 @@ def __init__(
         column: str | None = None,
         check_args: dict | None = None,
         min_dots: int = 5,
-        disp_rows: int = 10,
+        disp_rows: int = 20,
     ):
         self.failed_rows = failed_rows
         self.column = column
@@ -138,7 +152,7 @@ def log_one_liner(self, width: int):
 
     def failures_str(self) -> str:
         # These are the rows we want to display to the user.
-        disp = str(self.failed_rows.tail(self.disp_rows))
+        disp = df2str(self.failed_rows, max_rows=self.disp_rows)
         title = self.mini_signature()
         subtitle = f"Showing up to {self.disp_rows} rows"
 
diff --git a/tests/fixtures/frames.py b/tests/fixtures/frames.py
index 6954308..312616d 100644
--- a/tests/fixtures/frames.py
+++ b/tests/fixtures/frames.py
@@ -58,6 +58,26 @@ def abs_tol_frames():
     return baseline, candidate
 
 
+# Larger frames designed for tolerance sorting tests
+@pytest.fixture
+def abs_tol_frames_large():
+    dates = pd.date_range(start="2024-01-01", periods=1000, freq="D")
+    values = list(range(1000))
+    baseline = pd.DataFrame(
+        {
+            "date": dates,
+            "B": values,
+        }
+    ).set_index("date")
+    candidate = pd.DataFrame(
+        {
+            "date": dates,
+            "B": values[::-1],
+        }
+    ).set_index("date")
+    return baseline, candidate
+
+
 @pytest.fixture
 def equal_nan_frames():
     baseline = pd.DataFrame({"A": [1.0, None]})
diff --git a/tests/integration/test_rec.py b/tests/integration/test_rec.py
index 6a0b127..14e996b 100644
--- a/tests/integration/test_rec.py
+++ b/tests/integration/test_rec.py
@@ -79,3 +79,17 @@ def test_check_missin_indices(multi_index_frames):
 
     # Should pass because we are not checking for missing indices
     assert result.passed()
+
+
+def test_summary_runs_without_error(abs_tol_frames_large):
+    b, c = abs_tol_frames_large
+    rec = Rec(
+        columns={
+            "B": AbsTolCheck(tol=0.5, sort="desc"),
+        },
+        align_date_col="date",
+    )
+    result = rec.run(b, c)
+
+    assert not result.passed()
+    result.summary()