Reject ambiguous mapping CSV rows

sqlite_utils/export.py

@@ -1,4 +1,5 @@
 import csv
+import itertools
 from collections.abc import Mapping
 from typing import Any, Sequence, TextIO, Type, Union
 
@@ -26,15 +27,22 @@ def rows_to_csv_file(
         raise ValueError("Cursor does not have result columns")
 
     columns = [column[0] for column in description]
+    row_iterator = iter(cursor)
+    sentinel = object()
+    first_row = next(row_iterator, sentinel)
+    if isinstance(first_row, Mapping) and len(set(columns)) != len(columns):
+        raise ValueError("Mapping rows cannot represent duplicate column names")
+
     writer = csv.writer(file, dialect=dialect, **writer_kwargs)
     if header:
         writer.writerow(columns)
 
-    def rows():
-        for row in cursor:
-            if isinstance(row, Mapping):
-                yield [row[column] for column in columns]
-            else:
-                yield row
+    def normalize_row(row):
+        if isinstance(row, Mapping):
+            return [row[column] for column in columns]
+        return row
 
-    writer.writerows(rows())
+    if first_row is not sentinel:
+        writer.writerows(
+            normalize_row(row) for row in itertools.chain((first_row,), row_iterator)
+        )

tests/test_export.py

@@ -61,10 +61,7 @@ def test_rows_to_csv_file_with_mapping_row_factory():
     )
 
     def dict_factory(cursor, row):
-        return {
-            column[0]: value
-            for column, value in zip(cursor.description, row)
-        }
+        return {column[0]: value for column, value in zip(cursor.description, row)}
 
     db.row_factory = dict_factory
     cursor = db.execute("select id, name from creatures order by id")
@@ -74,6 +71,25 @@ def test_rows_to_csv_file_with_mapping_row_factory():
     db.close()
 
 
+def test_rows_to_csv_file_rejects_duplicate_mapping_columns():
+    db = sqlite3.connect(":memory:")
+    db.execute("create table creatures (id integer, name text)")
+    db.execute("insert into creatures (id, name) values (1, 'Cleo')")
+
+    def dict_factory(cursor, row):
+        return {column[0]: value for column, value in zip(cursor.description, row)}
+
+    db.row_factory = dict_factory
+    cursor = db.execute("select id as value, name as value from creatures")
+    output = io.StringIO(newline="")
+    with pytest.raises(
+        ValueError, match="Mapping rows cannot represent duplicate column names"
+    ):
+        rows_to_csv_file(cursor, output, lineterminator="\n")
+    assert output.getvalue() == ""
+    db.close()
+
+
 @pytest.mark.parametrize("header", (True, False))
 def test_rows_to_csv_file_requires_result_columns(header):
     db = sqlite3.connect(":memory:")