BUGs fixed

• Quote escape configured to double quote (quote value) character if escape not detected #414

• Delimiter detection returns first candidate delimiter even if it does not exist in the file #415

• context.getSelectedHeaders() in RowProcessor processStarted() can return invalid results #416

• DefaultNullRead of @Parsed does not work with enums #427

• Missing fields not initialized if nested beans present #432

• Possible race condition #424

• implicit limitation on max column name length? #438

Enhancements

• Delimiter detection returns first candidate delimiter even if it does not exist in the file #415

• Custom CsvFormatDetector #434

• Detects "whitespace" as delimiter instead of "comma" #420

BUGs fixed

• CSV auto-detection assigning line ending as quote escape (#409)

• FixedWidthFields.keepPadding not working (#405)

• Multi-char delimiter incorrectly detected inside quoted string (#404)

• Fixed the repeatable conversions initialization in the DefaultConversionProcessor (#399)

• Fix NPE in EnumConversion (#391)

• fixed quoted parser when using non-printable char as delimiter (#378)

Enhancements

• make the maxRowSample for parameter public configurable for CSV auto-detection (#408)

• settings.excludeFields() doesn't throw errors for a non-existing fields anymore. (#383)

• Expose InputAnalysisProcess implementations publicly (#381)

• add "com.googlecode.openbeans" as an optional OSGi dependency (#411)

BUGS FIXED:

• Value which contains line separator is NOT enclosed in quotes if line separator is 2 characters (#357)

• Headers in Record context are changed if parser instance is reused (#350)

• Record.getString(column) fails if type of column is declared as int and content in file is null (#355)

ENHANCEMENTS:

• Rows starting with comment char should be automatically quoted (#362)

BUGS:

• #345 parseLine() is throwing ClassCastException when using lookahead

• #337 Inconsistent parsing behavior when max. characters per column is set to -1 (unlimited)

• #343 FixedWidthParser drops first char of next record if last field of current record is empty

• #336 Single column, empty row CSV files result in empty rows and truncation of the last row

• #328 Auto detect format with upper and lower thom is not working

• fixed issue writing rows with selection of fields where the number of columns is larger than the selected/available fields.

ENHANCEMENTS:

• On CSV format auto-detection, use order of allowed delimiters as order of preference

• Performance improvement parsing values when maxCharsPerColumn = -1 (i.e. unlimited)

• added support for writing instances of Record directly.

BUGS:

• Headers being extracted when not expected, and this is leading to memory leak #326

• Unclear iteractions between iterate() methods and context/header parsing #314

• containsColumn throws ArrayIndexOutOfBoundsException, in combination with selectFields #316

• The result of the method getRecordNumber in TextParsingException is wrong. #324

ENHANCEMENTS:

• On CSV format auto-detection, resolve space as column separator #310

• Updated default excel serialization settings to ALWAYS escape either \r or \n as \r\n. #315

This release fixes bug reported in issue #309 - headers parsed from one input are used again when another input is parsed using the same parser instance.

Enhancements:

• Map column name to attribute #287: This is a big change. Basically you can now skip annotations altogether and manually define mappings from parsed columns to an object, including nested attributes.

Now you can call getColumnMapper() from:

1. any *Routines class
2. any *Bean*Processor class (including BeanWriterProcessor) for writing

    mapper.attributeToIndex("name", 0); //name goes to the first column
    mapper.attributeToColumnName("name", "client_name"); .// same thing, but the first column has header "client_name"

Nested classes are also supported, you can do stuff such as the following, where name is buried in an object structure (assume a class Purchase with a Buyer attribute, which in turn has a Contact attribute, where the name is located:

    mapper.methodToIndex("buyer.contact.name", 0); 
   
    // use methods too. This assumes Contact has a `fullName(java.lang.String)` setter method:
    mapper.methodToColumnName("buyer.contact.fullName", String.class, "client_name");

    // this maps getter and setter methods named `fullName` to column `client_name`. The getters or setters are used depending if you are writing to a file or reading from it.
    mapper.methodToColumnName("buyer.contact.fullName", "client_name");

You can also just give it a map:

    Map<String, Integer> mappings = new HashMap<String, Integer>();
    ... fill your map
    mapper.attributesToIndexes(mappings);

Contrived unit tests: https://github.com/uniVocity/univocity-parsers/blob/master/src/test/java/com/univocity/parsers/issues/github/Github_287.java

Other enhancements:

• Support CSV delimiters of more than one character #209: instead of using one character, delimiters now can be anything (e.g: ##, :-p etc).

• Support creation of immutable objects #280: create instances of java beans with private constructors and no setter methods.

• Enable case-sensitive column header matching #283: You can now target files with columns such as 'A', 'a', ' A ' and ' a '. Any existing code should not be affected. For example: if you use parserSettings.selectFields("a") and the parsed header is ' A ', then ' A ' will be selected as usual. This update allows one to use parserSettings.selectFields(" A ", "a") when headers such as ' A ' and 'a' are present (you can go wild here and target many different columns named aaa with different number of surrounding spaces or different character case combinations).

• CsvParser beginParsing closes the stream #303: Introduced flag to prevent the parser from automatically closing the input: parserSettings.setAutoClosingEnabled(false)

• Add option on FixedWidthParserSettings to keep padding of parsed values #276: the @FixedWidth annotation now has a keepPading property. This can also be set on the FixedWidthFields, using methods keepPadingOn(columns) and stripPaddingFrom(columns).

• Introduce UnescapedQuoteHandling.BACK_TO_DELIMITER #259: if an unescaped quote is detected, the parser will re-process the parsed value and split the unescaped value into individual values after each delimiter found.

Bugfixes:

• An extra row with null returned #306

• Cannot determine column separator #305

• Wrong line ending detection when normalizeLineEndingsWithinQuotes = false #299

• Column selection makes @Validate annotation misbehave #296

• Fixed width parsing with look-ahead and non-contiguous field definitions #294

• CsvParser.parse ignores headers parser setting in processStarted Processor's method #289

Enhancements:

• allowing users to subclass ValidatedConversion
• better handling of InvocationTargetExceptions when dealing with errors thrown from getters/setters

Bugfixes:

• custom validations weren't working when writing
• CSV autodetection failed in some cases (issue #272)

Enhancements:

• Performance improvements skipping values of de-selected fields with the CSV parser.

• Adding support for regex validation and custom validations on class attributes and methods annotated with @Validate

Bugfixes:

• CsvRoutines.getInputDimension() returns one row less rowCount regardless of csvParserSettings.setHeaderExtractionEnabled() #262

• record.toFieldMap() not working on FixedWidth #258

Bugfixes:

• Problems when using two BeanProcessors with an InputValueSwitch #256

• Column names in the header row get printed in wrong order despite index values being set correctly #255

• Inconsistent result of parseLine on empty set of selected indexes #250

Fixed a couple of bugs:

• Annotated fields won't be processed when reading rows with less elements (#253)

• Incorrect handling of columnReorderingEnabled = false ( #254)

Enhancements:

• Introduce @Validate annotation and conversion to perform basic validations #251

Bugfixes:

• Inconsistent result of parseLine on empty set of selected indexes #250

• Fixed comments collecting on buffer update #252

Bugfixes:

• Support Enum alternative values #240

• better error message for more than max columns #247

Enhancements:

• FixedWidthWriter does not honour FixedWidthWriterSettings.setNullValue #238

remove dependency on internal JDK's "memberValues" field used when processing @Copy annotations.

Bug fixes

• Concurrency issue when calling stopParsing() (#231)
• Error deriving header names from annotated java beans when writing objects with meta-annotations.

Enhancements

• Implemented trim quoted values support (#230)

Issue #228 revealed a regression bug on the CSV parser which was introduced in version 2.6.0

The last column of a row would be ignored if it's empty and preceded by a quoted value, e.g.

A row such as "A", would be parsed as [A] instead of [A, null]

This new release fixes the bug.

Enhacements:

• CSV parser now parses quoted values ~30% faster

• CSV format detection process has option provide a list of possible delimiters, in order of priority ( i.e. settings.detectFormatAutomatically( '-', '.');) (#214 )

• CSV writer allows selecting columns that should always have quotes (i.e. writerSettings.quoteFields("C", "D"); or writerSettings.quoteIndexes(0, 1, 4, 5);) (#191)

• Introduced support for selecting and aggregating values in lists of java beans when inputs have repeated header names (#188)

• Context class has two new methods for facilitating the usage of Record inside RowProcessor (#211):

• Adjusting method signatures of AbstractWriter to properly handle writing rows based on collections of objects of any type (from a StackOverflow complaint).

public Record toRecord(String[] row);

public RecordMetaData recordMetaData();

Bugfixes:

• NullPointer when stopping parser when nothing is parsed (#219)

• Reusing CsvRoutines object results in unexpected output (#224)

Bugfixes

• #205 OOM on skip lines. AbstractCharInputReader.skipLines

• #212 Bad results if charset is null in beginParsing

• #203 CsvRoutines generating empty output with keepResourcesOpen

• Fixed incorrect behavior processing comment lines at the end of the input, when no line ending is present after the commented out line.

ENHANCEMENTS

• Introduced the CompositeProcessor based on question raised in ticket #206

• Adjustment on CsvFormatDetector to improve chances of detecting delimiters in small inputs.

• Added expectedRowCount parameter to all methods that produce lists of rows, records or beans to prevent slow reallocation operations inside the resulting ArrayList. Useful when processing large inputs.

• Trailing spaces in csv line leads to null instead of empty String (#196)
• On CSV, keepQuotes fail to print out closing quote if it's followed by a whitespace and a line ending (#197)
• CsvParser does not properly detect delimiter (with setDelimiterDetectionEnabled) (#198)
• Routines' keepResourcesOpen not respected with writeAll(list of objects) (#201)

This version was released to bury the intermittent results produced by the parser when processing files and input streams without an explicit character encoding while readInputOnSeparateThread=true as reported on issue #194

This problem appeared on version 2.5.0 onward and has been plaguing users until version 2.5.6, and is finally gone for good.

This is the first Java 9 compatible version

DO NOT use any previous release with Java 9 as it will blow up.

• string.getChars used to throw ArrayIndexOutOfBoundsException (tested on JDK 6-8) but now it throws StringIndexOutOfBoundsException. This version was updated to catch IndexOutOfBoundsException instead so it can work with all JDKs (tested from 6 to 9).

• SimpleDateFormat - Changes in the locales "broke" previously working date formats. If the default locale is "en_AU" then the following won't work with JDK 9: new SimpleDateFormat("yyyy-MMM-dd").parse("2015-DEC-25")

Classes with annotations for dates now allow users to provide a locale. For example:

    @Parsed
    @Format(formats = "dd-MMM-yyyy", options = "locale=en_US_WIN")
    private Date myDate;

    //if you are in Australia, MMM will translate to the abbreviated format followed by period, 
    //i.e. "October"  becomes "Oct.". The formatter won't work to parse dates such as "2015-Oct-20"
    //Use "locale=en" to make it behave as it always did in Java 8 or earlier.
    @Parsed
    @Format(formats = "dd-MMM-yyyy", options = "locale=en")
    private Date australianDay;

Also fixed a bug on the TrimConversion introduced in version 2.5.5 - it would break processing blank or empty strings.

Fixed concurrency issue processing File or InputStream with no explicit character encoding provided, when setReadInputOnSeparateThread=true (the default). The issue could make the parser discard the first character of the input.

More details in: https://github.com/uniVocity/univocity-parsers/issues/186 and https://github.com/uniVocity/univocity-parsers/issues/187

** Other adjustments **

• Improved CSV format detection

• Fixed TrimConversion to remove trailing whitespaces when input is truncated to a maximum length.

BUGFIXES

Invalid values being produced when column selection is defined, reordering is disabled and number of headers is less than number of selected columns. (https://github.com/uniVocity/univocity-parsers/issues/183)

ParsingContext of *Routines returns information of the following row instead of the current (https://github.com/uniVocity/univocity-parsers/issues/184)

currentParsedContent returns a first-char-repeated string from ByteArrayInputStream with setReadInputOnSeparateThread disabled (https://github.com/uniVocity/univocity-parsers/issues/185)

Fixed issues handling values with unescaped quotes at the end of the input, and in combination with the keepQuotes flag - commit.

ENHANCEMENTS

Allowing the same column to feed data into multiple fields of a java bean and its nested fields. Validation kicks in only if writing.

Bug fixes

• CsvParserSettings.setDelimiterDetectionEnabled() not working with input file/stream when the character encoding is not explicitly provided. (https://github.com/uniVocity/univocity-parsers/issues/178)

• CSV format autodetection thrown off by fields with single or double quote in the middle of the value (also in https://github.com/uniVocity/univocity-parsers/issues/178)

• Fixed-width header extraction fails (https://github.com/uniVocity/univocity-parsers/issues/182)

• Failure to extract header row with FixedWidthParser + annotations (https://github.com/uniVocity/univocity-parsers/issues/180)

Enhancements

• Custom conversion classes can now be defined without a String[] args constructor (https://github.com/uniVocity/univocity-parsers/issues/181) if no parameters are used.

Fixed bug in CSV parser - line ending following escaped quote was handled was end of the record (Issue https://github.com/uniVocity/univocity-parsers/issues/177)

The introduction of support for BOM marker handling created an unwanted side effect when the user does not provide the character encoding of the file being processed: performance is reduced significantly

This version fixes the problem and the performance should be equivalent of previous versions, regardless of the user providing a character encoding explicitly, or the input having a BOM marker.

Relevant issue: https://github.com/uniVocity/univocity-parsers/issues/176

ENHANCEMENTS

• Added methods to the parser that return Iterable<String[]> or Iterable(https://github.com/uniVocity/univocity-parsers/issues/151)

Rows can now be iterated like this:

	for (String[] row : parser.iterate(input, "UTF-8")) {
		//do stuff
	}

And records:

	for (Record row : parser.iterateRecords(new FileInputStream(input), "UTF-8")) {
		//do stuff
	}

• Support multiple header names in field attribute of @Parsed annotation (https://github.com/uniVocity/univocity-parsers/issues/150)

You can now use the same POJO to parse different files with different headers.

• Adding a HeaderTransformer to the @Nested annotation (https://github.com/uniVocity/univocity-parsers/issues/159)

This allows one to nest the same class inside a parent multiple times, by changing the names assigned to the nested properties. Example for a car with 4 wheels:

	public static class Wheel {
		@Parsed
		String brand;

		@Parsed
		int miles;
	}

public static class Car {
		@Nested
		Wheel frontLeft;

		@Nested
		Wheel frontRight;

		@Nested
		Wheel rearLeft;

		@Nested
		Wheel rearRight;
}

To parse an input with organized like this: frontLeftWheelBrand,frontLeftWheelMiles,frontRightWheelBrand,frontRightWheelMiles,rearLeftWheelBrand,rearLeftWheelMiles,rearRightWheelBrand,rearRightWheelMiles, we need to apply a transformation to the property names of each Wheel instance. This can be done with something like this:

	public static class NameTransformer extends HeaderTransformer {

		private String prefix;

		public NameTransformer(String... args) {
			prefix = args[0];
		}

		@Override
		public String transformName(Field field, String name) {
			return prefix + Character.toUpperCase(name.charAt(0)) + name.substring(1);
		}
	}

Now we can change the Nested annotations to be:

public static class Car {
		@Nested(headerTransformer = NameTransformer.class, args = "frontLeftWheel")
		Wheel frontLeft;

		@Nested(headerTransformer = NameTransformer.class, args = "frontRightWheel")
		Wheel frontRight;

		@Nested(headerTransformer = NameTransformer.class, args = "rearLeftWheel")
		Wheel rearLeft;

		@Nested(headerTransformer = NameTransformer.class, args = "rearRightWheel")
		Wheel rearRight;
}

And parse the input with:

	List<Car> cars = new CsvRoutines(settings).parseAll(Car.class, input);

• added support for annotations in methods (https://github.com/uniVocity/univocity-parsers/issues/160)

@Parsed or @Nested annotations now work on individual methods. Modifying the Car example above to use a List<Wheel>, one can do the followng:

	public static class Car {
		List<Wheel> wheels = new ArrayList<Wheel>();
		
		@Nested(headerTransformer = NameTransformer.class, args = "frontLeftWheel")
		private void setWheel1(Wheel wheel) { //bad method name created on purpose. Also private.
			wheels.add(wheel);
		}

		@Nested(headerTransformer = NameTransformer.class, args = "frontRightWheel")
		private void setWheel2(Wheel wheel) { //bad method name created on purpose. Also private.
			wheels.add(wheel);
		}

		@Nested(headerTransformer = NameTransformer.class, args = "rearLeftWheel")
		private void setWheel3(Wheel wheel) { //bad method name created on purpose. Also private.
			wheels.add(wheel);
		}

		@Nested(headerTransformer = NameTransformer.class, args = "rearRightWheel")
		private void setWheel4(Wheel wheel) { //bad method name created on purpose. Also private.
			wheels.add(wheel);
		}
	}

• added support and automatic handling of files with a BOM marker (https://github.com/uniVocity/univocity-parsers/issues/154)

• Introduced properties from and to to the @FixedWidth annotation, along with FixedWidthFields.addField(int startPosition, int endPosition) to allow declaring field ranges instead of fixed lengths (https://github.com/uniVocity/univocity-parsers/issues/166)

• enabled column reordering in CSV, TSV and Fixed-Width writers. This allows users to select which columns of their input should be written and prevents generating empty columns. Just use selectFields with setColumnReorderingEnabled(true) (https://github.com/uniVocity/univocity-parsers/issues/167)

• added a keepResourcesOpen property in CsvRoutines, TsvRoutines and FixedWidthRoutines to control whether or not to close the Writer after writing data to the output. Also controls whether the ResultSet will closed when dumping data from it. (https://github.com/uniVocity/univocity-parsers/issues/172)

BUGS FIXED

• CSV format auto-detection thrown off by seemingly regular CSV (https://github.com/uniVocity/univocity-parsers/issues/161)

• Can't provide custom headers when dumping a ResultSet (https://github.com/uniVocity/univocity-parsers/issues/157)

• ParsingContext.currentParsedContent returns null when input is single length string (https://github.com/uniVocity/univocity-parsers/issues/165)

• NullValue and EmptyValue are both applied (https://github.com/uniVocity/univocity-parsers/issues/158)

• NumberFormatException when using record.getDate("field") in a concurrent environment (https://github.com/uniVocity/univocity-parsers/issues/156)

• AutomaticConfiguration does not work with MultiBeanListProcessor (https://github.com/uniVocity/univocity-parsers/issues/149)

This release includes:

• a fix for concurrency issues processing annotations: https://github.com/uniVocity/univocity-parsers/issues/146

• internal adjustments for exception handling and generation of error messages.

Enhancements

• Added support for nested objects with the newly introduced @Nested annotation: https://github.com/uniVocity/univocity-parsers/issues/139
• CsvRoutines and other routine classes provide a ParsingContext object: https://github.com/uniVocity/univocity-parsers/issues/136

Bugfixes:

• Fixed incorrect escape of CSV when writing a quote escape character with setQuoteEscapingEnabled=true before writing any other character that would prompt the writer to enclose the field in quotes: https://github.com/uniVocity/univocity-parsers/issues/143
• Fixed handling of unquoted values where two consecutive unescaped quotes appeared in the input: https://github.com/uniVocity/univocity-parsers/issues/143
• IndexOutOfBoundsException when processing index-based annotations: https://github.com/uniVocity/univocity-parsers/commit/33a099c57a61d14d2fb2a58c22cd84806e9d27c8
• Fixed width writing from ResultSet should respect user provided field length instead of using the resultset length - https://github.com/uniVocity/univocity-parsers/issues/135
• Parser did not throw error in case input fails to be read (e.g. in case of socket timeouts, etc) and would just silently stop. https://github.com/uniVocity/univocity-parsers/issues/140

This release includes a couple of bug fixes:

• setColumnReorderingEnabled(false) not working when parsing single line: https://github.com/uniVocity/univocity-parsers/issues/131
• index property of @Parsed annotation not handled correctly: https://github.com/uniVocity/univocity-parsers/issues/132

...and some internal refactoring that should not affect anyone.