v3.6.0

Added

• Configuration of maximum fields, maximum field size, and maximum record size via record handler
• Fluent configuration for CsvRecordHandler, NamedCsvRecordHandler, and StringArrayHandler
• maxBufferSize(int) in CsvReaderBuilder and IndexedCsvReaderBuilder to alter the maximum buffer size of the parser

Changed

• Use ReentrantLock in IndexedCsvReader instead of synchronized to prevent pinning of virtual threads
• More precise error messages when exceeding the maximum field or record size
• More precise error when parsing error occurs within IndexedCsvReader
• Apply length constraints (maximum field size and maximum record size) after applying field modifiers

Deprecated

• Setting the maximum field size (and maximum buffer size) via system property fastcsv.max.field.size
• Setting the maximum field count per record via system property fastcsv.max.field.count
• Constructor initialization of CsvRecordHandler, NamedCsvRecordHandler, and StringArrayHandler

v3.5.0

Added

• Support InputStream as an input source for CsvReader including BOM detection #130
• Support OutputStream as an output target for CsvWriter

v3.4.0

Added

• Add skipLines(int) and skipLines(Predicate<String>, int) to CsvReader to skip lines before the actual CSV data starts

v3.3.1

Fixed

• Fixed a bug in CsvReader where lines were mistakenly treated as empty and skipped when skipEmptyLines was set (default). These affected lines made up solely of field separators, solely empty quoted fields, or fields rendered empty after applying optional field modifiers.

v3.3.0

Added

• Implement Flushable interface for CsvWriter to allow flushing the underlying writer
• Implement autoFlush option for CsvWriter to automatically flush the writer after writing a record
• Implement toConsole method for CsvWriter to write records to the console

v3.2.0

Added

• Add writeRecord() to CsvWriter to allow writing records field by field
• Allow overwriting the limits of 16K fields per record and 16M characters per field (#104); Thanks to @Obolrom!

v3.1.0

Added

• Add acceptCharsAfterQuotes() to CsvReaderBuilder and IndexedReaderBuilder to allow even stricter parsing

Changed

• Improved reader performance

v3.0.0

Added

• IndexedCsvReader for random access to CSV files
• FieldModifier for modifying fields while reading CSV files
• Allow custom quote strategies for CsvWriter
• CsvCallbackHandler for more flexible usage of CsvReader
• Support for optional BOM header when reading CSV files
• Method NamedCsvRecord.findField for optional field access
• Allow READ comment strategy for CSV data with a header
• Metadata for OSGi capability

Changed

• Updated from Java 8 to Java 11
• Updated naming (rows/lines -> records, columns -> fields, differentiate between lines and records)
• NamedCsvReader replaced by CsvReader.builder().ofNamedCsvRecord()
• build methods in CsvReaderBuilder with callback handlers and ofCsvRecord / ofNamedCsvRecord as convenience methods
• Rename errorOnDifferentFieldCount() to ignoreDifferentFieldCount()
• QuoteStrategy is now an interface – defaults are provided by QuoteStrategies
• Throw CsvParseException instead of IOException when maximum field size is exceeded
• NamedCsvRecord extends CsvRecord and provides more access methods
• Raised the maximum field size to 16 MiB to match SUPER data type capabilities of Amazon Redshift
• Limit the maximum field count per record to 16,384 to prevent OutOfMemoryErrors
• Limit the maximum record size to 64 MiB to prevent OutOfMemoryErrors
• Several performance improvements
• Improved documentation and error messages

Removed

• Removed isEmpty() in CsvRecord as it was formerly only used for skipping empty records

Fixed

• Do not throw an exception when reading comments while enabling different field count checking

v2.2.2

Added

• New quote strategy that adds quotes only for non-empty fields #80

v2.2.1

Fixed

• Fixed a problem when refilling the input buffer while parsing nonconforming data (quote character within unquoted field) #67