fix: improve CSV parser error handling

test/mv/membership/import/csv_parser_test.exs

@@ -109,7 +109,7 @@ defmodule Mv.Membership.Import.CsvParserTest do
       assert {:ok, headers, rows} = CsvParser.parse(csv_content)
 
       assert headers == ["email"]
-      # Lines 2, 3, 4 are empty (skipped), line 4 has data
+      # Lines 2 & 3 are empty (skipped), line 4 has data
       assert rows == [{4, ["john@example.com"]}]
     end
   end
@@ -156,6 +156,19 @@ defmodule Mv.Membership.Import.CsvParserTest do
       assert headers == ["email", "name"]
       assert rows == [{2, ["john@example.com", "John \"Johnny\" Doe"]}]
     end
+
+    test "handles multiline quoted fields with correct line numbering" do
+      # Header line 1
+      # Data record starts line 2, contains "foo\nbar" in a field
+      # Record ends physically at line 3
+      # Expected: row gets line number 2 (start line)
+      csv_content = "email;description\njohn@example.com;\"foo\nbar\""
+
+      assert {:ok, headers, rows} = CsvParser.parse(csv_content)
+
+      assert headers == ["email", "description"]
+      assert rows == [{2, ["john@example.com", "foo\nbar"]}]
+    end
   end
 
   describe "error handling" do
@@ -170,12 +183,23 @@ defmodule Mv.Membership.Import.CsvParserTest do
       assert reason =~ "CSV file is empty"
     end
 
-    test "returns {:error, reason} for invalid CSV format" do
-      # Unbalanced quotes
-      csv_content = "email;name\n\"john@example.com;John"
+    test "returns {:error, reason} for invalid UTF-8 content" do
+      # Invalid UTF-8 sequence
+      invalid_utf8 = <<0xFF, 0xFE, 0xFD>>
+
+      assert {:error, reason} = CsvParser.parse(invalid_utf8)
+      assert reason =~ "UTF-8"
+    end
+
+    test "returns {:error, reason} for unparsable data row" do
+      # Malformed CSV row that cannot be parsed
+      # NimbleCSV will throw an exception for unclosed quotes
+      csv_content = "email;name\njohn@example.com;\"unclosed quote"
 
       assert {:error, reason} = CsvParser.parse(csv_content)
       assert is_binary(reason)
+      # Error message should indicate parsing failure
+      assert reason =~ "parse" or reason =~ "CSV"
     end
   end