lib/mv/membership/import/csv_parser.ex

@@ -16,6 +16,7 @@ defmodule Mv.Membership.Import.CsvParser do
 
   @utf8_bom <<0xEF, 0xBB, 0xBF>>
   @quote ?"
+  @max_error_snippet_length 50
 
   @type line_number :: pos_integer()
   @type row :: [String.t()]
@@ -27,11 +28,8 @@ defmodule Mv.Membership.Import.CsvParser do
          content <- file_content |> strip_bom() |> normalize_line_endings(),
          :ok <- validate_content_not_empty(content),
          {:ok, header_record, data_records} <- extract_header_and_data(content),
-         :ok <- validate_header_not_empty(header_record),
+         :ok <- validate_header_not_empty(header_record) do
-         {:ok, headers, rows} <- parse_csv_records(header_record, data_records) do
+      parse_csv_records(header_record, data_records)
-      {:ok, headers, rows}
-    else
-      {:error, reason} -> {:error, reason}
     end
   end
 
@@ -55,7 +53,8 @@ defmodule Mv.Membership.Import.CsvParser do
     end
   end
 
-  @spec extract_header_and_data(binary()) :: {:ok, binary(), [{line_number(), binary()}]} | {:error, String.t()}
+  @spec extract_header_and_data(binary()) ::
+          {:ok, binary(), [{line_number(), binary()}]} | {:error, String.t()}
   defp extract_header_and_data(content) do
     records = split_records_with_line_numbers(content)
 
@@ -126,11 +125,10 @@ defmodule Mv.Membership.Import.CsvParser do
   end
 
   # Parses exactly one record (string without trailing newline is fine).
-  # Returns {:ok, row} or {:error, reason}.
+  # Returns `{:ok, row}` or `{:error, reason}`.
   @spec parse_single_record(module(), binary(), String.t() | nil) ::
           {:ok, row()} | {:error, String.t()}
   defp parse_single_record(parser, record, error_reason_if_empty) do
-    try do
     # NimbleCSV is happiest if there's a newline at the end.
     rows = parser.parse_string(ensure_trailing_newline(record), skip_headers: false)
 
@@ -147,7 +145,6 @@ defmodule Mv.Membership.Import.CsvParser do
     e ->
       {:error, "Failed to parse CSV: #{Exception.message(e)}"}
   end
-  end
 
   @spec ensure_trailing_newline(binary()) :: binary()
   defp ensure_trailing_newline(str) do
@@ -155,19 +152,43 @@ defmodule Mv.Membership.Import.CsvParser do
   end
 
   # --- Data parsing preserving *physical* line numbers ---
-
+  #
+  # Parses data records while preserving physical line numbers.
+  # Skips empty rows but maintains correct line numbering for error reporting.
+  #
   @spec parse_data_records(module(), [{line_number(), binary()}]) ::
           {:ok, [numbered_row()]} | {:error, String.t()}
   defp parse_data_records(parser, data_records) do
-    try do
     rows =
       data_records
       |> Enum.reduce_while([], fn {line_no, record}, acc ->
-          case String.trim(record) do
+        process_data_record(parser, line_no, record, acc)
-            "" ->
+      end)
-              {:cont, acc}
 
-            _ ->
+    case rows do
+      {:error, reason} -> {:error, reason}
+      rows -> {:ok, Enum.reverse(rows)}
+    end
+  rescue
+    e ->
+      {:error, "Failed to parse CSV: #{Exception.message(e)}"}
+  end
+
+  @spec process_data_record(module(), line_number(), binary(), [numbered_row()]) ::
+          {:cont, [numbered_row()]} | {:halt, {:error, String.t()}}
+  defp process_data_record(parser, line_no, record, acc) do
+    trimmed = String.trim(record)
+
+    if trimmed == "" do
+      {:cont, acc}
+    else
+      process_non_empty_record(parser, line_no, record, acc)
+    end
+  end
+
+  @spec process_non_empty_record(module(), line_number(), binary(), [numbered_row()]) ::
+          {:cont, [numbered_row()]} | {:halt, {:error, String.t()}}
+  defp process_non_empty_record(parser, line_no, record, acc) do
     parsed = parser.parse_string(ensure_trailing_newline(record), skip_headers: false)
 
     case parsed do
@@ -180,19 +201,10 @@ defmodule Mv.Membership.Import.CsvParser do
 
       # unparsable row -> return error with line number
       _ ->
-                  snippet = String.slice(record, 0, min(50, String.length(record)))
+        snippet =
-                  {:halt, {:error, "Failed to parse CSV data at line #{line_no}: #{inspect(snippet)}"}}
+          String.slice(record, 0, min(@max_error_snippet_length, String.length(record)))
-              end
-          end
-        end)
 
-      case rows do
+        {:halt, {:error, "Failed to parse CSV data at line #{line_no}: #{inspect(snippet)}"}}
-        {:error, reason} -> {:error, reason}
-        rows -> {:ok, Enum.reverse(rows)}
-      end
-    rescue
-      e ->
-        {:error, "Failed to parse CSV: #{Exception.message(e)}"}
     end
   end
 
@@ -203,8 +215,8 @@ defmodule Mv.Membership.Import.CsvParser do
 
   # --- Record splitting with correct line numbers (quote-aware) ---
   #
-  # This splits the CSV into "records" separated by newline *outside quotes*,
+  # Splits the CSV into records separated by newline *outside quotes*.
-  # returning [{start_line_number, record_string_without_newline}, ...]
+  # Returns `[{start_line_number, record_string_without_newline}, ...]`.
   #
   # Line numbers are 1-based and represent the physical line in the CSV file.
   # Empty lines are included in the numbering (they're just skipped later).
@@ -228,7 +240,21 @@ defmodule Mv.Membership.Import.CsvParser do
     Enum.reverse(acc)
   end
 
-  # EOF
+  # Recursively splits CSV content into records with correct line numbering.
+  #
+  # Handles quote-aware parsing:
+  # - Escaped quotes (`""`) inside quoted fields are preserved
+  # - Newlines inside quotes are part of the record but advance line counter
+  # - Newlines outside quotes end a record
+  #
+  # Parameters:
+  # - `content` - Remaining binary content to parse
+  # - `acc` - Accumulated records `[{line_number, record}, ...]`
+  # - `buf` - Current record buffer (reversed byte list)
+  # - `in_quotes` - Whether we're currently inside a quoted field
+  # - `line` - Current physical line number
+  # - `start_line` - Line number where current record started
+  #
   @spec do_split(
           binary(),
           [{line_number(), binary()}],