Plot error message (#441)

* success case

* Plot error message

testcase

Fix

Fix

Fix

test fix

test fix

test fix

changelog

usageexception

test

test

error

error

error

error

error

changed to runtimeerror

quest db test

Revert test

Revert test

exception type test

Review comments

changelog

rebase

changelog

Review comments

changed to sqlite

change sequence

lint

revert

added tmp_empty

failure case

CHANGELOG.md

@@ -4,6 +4,7 @@
 * [Feature] Using native DuckDB `.df()` method when using `autopandas` 
 
 * [Doc] documenting `%sqlcmd tables`/`%sqlcmd columns`
+* [Feature] Better error messages when function used in plotting API unsupported by DB driver (#159)
 
 ## 0.7.4 (2023-04-28)
 No changes

src/sql/plot.py

@@ -6,6 +6,8 @@
 from jinja2 import Template
 
 from sql.util import flatten
+from sqlalchemy.exc import ProgrammingError
+from sql import exceptions
 
 try:
     import matplotlib.pyplot as plt
@@ -30,6 +32,7 @@ def _summary_stats(conn, table, column, with_=None):
 
     if not conn:
         conn = sql.connection.Connection.current
+    driver = conn._get_curr_sqlalchemy_connection_info()["driver"]
 
     template = Template(
         """
@@ -44,7 +47,16 @@ def _summary_stats(conn, table, column, with_=None):
 
     query = template.render(table=table, column=column)
 
-    values = conn.execute(query, with_).fetchone()
+    try:
+        values = conn.execute(query, with_).fetchone()
+    except ProgrammingError as e:
+        print(e)
+        raise exceptions.RuntimeError(
+            f"\nEnsure that percentile_disc function is available on {driver}."
+        )
+    except Exception as e:
+        raise e
+
     keys = ["q1", "med", "q3", "mean", "N"]
     return {k: float(v) for k, v in zip(keys, flatten(values))}
 

src/tests/integration/test_mssql.py

@@ -1,6 +1,7 @@
 import pyodbc
 import pytest
 from matplotlib import pyplot as plt
+from IPython.core.error import UsageError
 
 
 def test_query_count(ip_with_MSSQL, test_table_name_dict):
@@ -112,3 +113,17 @@ def test_sqlplot_boxplot(ip_with_MSSQL, cell):
     out = ip_with_MSSQL.run_cell(cell)
 
     assert type(out.result).__name__ in {"Axes", "AxesSubplot"}
+
+
+def test_unsupported_function(ip_with_MSSQL, test_table_name_dict):
+    # clean current Axes
+    plt.cla()
+    out = ip_with_MSSQL.run_cell(
+        f"%sqlplot boxplot --table " f"{test_table_name_dict['taxi']} --column x"
+    )
+    assert isinstance(out.error_in_exec, UsageError)
+    assert "Ensure that percentile_disc function is available" in str(out.error_in_exec)
+    assert (
+        "If you need help solving this issue, "
+        "send us a message: https://ploomber.io/community" in str(out.error_in_exec)
+    )

src/tests/test_plot.py

@@ -7,6 +7,7 @@
 from sql.connection import Connection
 from pathlib import Path
 import pytest
+from sqlalchemy.exc import OperationalError
 
 
 class DictOfFloats(Mapping):
@@ -81,6 +82,36 @@ def test_boxplot_stats_exception(chinook_db, ip_empty):
         )
 
 
+def test_summary_stats(chinook_db, ip_empty, tmp_empty):
+    Path("data.csv").write_text(
+        """\
+x, y
+0, 0
+1, 1
+2, 2
+5, 7
+9, 9
+"""
+    )
+    ip_empty.run_cell("%sql duckdb://")
+    ip_empty.run_cell("%sql INSTALL 'sqlite_scanner';")
+    ip_empty.run_cell("%sql commit")
+    ip_empty.run_cell("%sql LOAD 'sqlite_scanner';")
+    result = plot._summary_stats(Connection.current, "data.csv", column="x")
+    expected = {"q1": 1.0, "med": 2.0, "q3": 5.0, "mean": 3.4, "N": 5.0}
+    assert result == expected
+
+
+def test_summary_stats_missing_file(chinook_db, ip_empty):
+    ip_empty.run_cell("%sql duckdb://")
+    ip_empty.run_cell("%sql INSTALL 'sqlite_scanner';")
+    ip_empty.run_cell("%sql commit")
+    ip_empty.run_cell("%sql LOAD 'sqlite_scanner';")
+    with pytest.raises(OperationalError) as e:
+        plot._summary_stats(Connection.current, "data.csv", column="x")
+    assert 'No files found that match the pattern "data.csv"' in str(e)
+
+
 @pytest.mark.parametrize(
     "cell, error_type, error_message",
     [