Read fields with a DataReader

ToDataReader returns the format’s concrete forward-only reader. ToDbDataReader adapts it to the standard ADO.NET DbDataReader. Standard interfaces do not expose every PocketCsvReader helper.

Obtaining an IDataReader

using var reader = csv.ToDataReader("customers.csv");
while (reader.Read())
    Console.WriteLine(reader.GetString(reader.GetOrdinal("Name")));

Create csv with the complete first-CSV example. The current row exists only after a successful Read; field counts and labels in changing labeled formats must be checked on that row. Dispose the reader even after an early exit. See stream ownership.

Reading Values with IDataReader

Iterating Over Records

Read advances synchronously. Use ReadAsync for asynchronous I/O and cancellation. Getters remain synchronous.

Accessing Typed Field Values

Use GetInt32, GetInt64, GetBoolean, GetDecimal, GetString, GetDateTime, or GetFieldValue<T> on the concrete reader. Access by ordinal or supported name overload. Getters parse on demand with field/runtime settings; a declared schema controls boxed values and reported types. See schemas and cultures.

GetString returns logical text after applicable format decoding/sanitization. GetRawString returns encoded/raw field text according to the format, including quote/escape syntax where retained. JSON/logfmt decoding differs from CSV sequence replacement; see value handling.

Generic and Custom Type Access

GetFieldValue<T>(ordinal, parser) and name overloads support string delegates or explicitly typed ReadOnlySpan<char> delegates. See custom parsers for complete definitions, precedence and lifetime rules. Null/reference typed reads can return default without invoking the delegate.

Accessing Boxed Values

GetValue, GetValues, and indexers return parsed boxed values using schema/runtime settings. Untyped fields remain strings. Boxing value types allocates; a span parser does not eliminate that result allocation.

Working with Arrays

GetArray<T>, GetArray, and GetArrayItem<T> expose array fields. Schema-declared one-dimensional arrays can also be returned by GetValue. See array schemas, including whole-array parser precedence and limitations.

Checking for Nulls

Use IsDBNull(index) or the concrete reader’s IsDBNull(name) for a present field. Use TryGetOrdinal first when a label might be absent; see optional fields.

Metadata Access

Member Meaning
GetName / GetOrdinal Current field name/index; ordinal case-sensitive matching
TryGetOrdinal Non-throwing absent-name lookup on concrete readers
GetFieldType Type determined by schema and reader behavior
GetDataTypeName Configured source type name
FieldCount Current reader shape; can vary for labeled records

Headerless CSV generates names such as field_0. Do not treat a field count/name lookup before the first successful advancement as a portable operation across formats.

Validating cells

The complete IsValid contract and malformed-input handling are in validation and errors.

Source positions

Enable tracking before reading; see Source positions for units and lifecycle. Generated row numbers are a separate output column.

Filter before reading

Configure reader filters on the builder to make Read() and ReadAsync() skip rejected rows before returning a current record. Conditions can inspect sanitized spans, parsed values, or several fields through a read-only row view. Compatible parsed values are reused by subsequent getters.


Choose an output model
Filter rows before projection