fix: auto-detect single-column CSV files

sqlite_utils/utils.py

@@ -389,14 +389,24 @@ class Format(enum.Enum):
             with buffered:
                 return rows_from_file(buffered, format=Format.JSON)
         else:
-            dialect = csv.Sniffer().sniff(
-                first_bytes.decode(encoding or "utf-8-sig", "ignore")
-            )
+            try:
+                dialect = csv.Sniffer().sniff(
+                    first_bytes.decode(encoding or "utf-8-sig", "ignore")
+                )
+            except csv.Error:
+                # Single-column files have no delimiter to sniff (e.g. a bare
+                # "id\n1\n2\n" list). Fall back to the default CSV dialect so
+                # DictReader still loads them as a single column.
+                dialect = None
             rows, _ = rows_from_file(
                 buffered, format=Format.CSV, dialect=dialect, encoding=encoding
             )
             # Make sure we return the format we detected
-            detected_format = Format.TSV if dialect.delimiter == "\t" else Format.CSV
+            detected_format = (
+                Format.TSV
+                if dialect is not None and dialect.delimiter == "\t"
+                else Format.CSV
+            )
             return (
                 _extra_key_strategy(
                     cast(Iterable[dict[str | None, object]], rows),

tests/test_utils.py

@@ -55,6 +55,21 @@ def test_hash_record():
     )
 
 
+@pytest.mark.parametrize(
+    "content,expected",
+    (
+        (b"apple\nbanana\ncherry\n", [{"apple": "banana"}, {"apple": "cherry"}]),
+        (b"id\n1\n2\n", [{"id": "1"}, {"id": "2"}]),
+    ),
+)
+def test_rows_from_file_detects_single_column_csv(content, expected):
+    # A single-column file has no delimiter for csv.Sniffer to detect; it
+    # used to raise csv.Error("Could not determine delimiter").
+    rows, format_ = utils.rows_from_file(io.BytesIO(content))
+    assert format_ == utils.Format.CSV
+    assert list(rows) == expected
+
+
 def test_maximize_csv_field_size_limit():
     # Reset to default in case other tests have changed it
     csv.field_size_limit(utils.ORIGINAL_CSV_FIELD_SIZE_LIMIT)