@Override
public void run(InputRow row, int distinctCount) {
for (InputColumn<String> column : _columns) {
String value = row.getValue(column);
StringAnalyzerColumnDelegate delegate = _columnDelegates.get(column);
delegate.run(row, value, distinctCount);
}
}
@Override
public StringAnalyzerResult getResult() {
logger.info("getResult()");
CrosstabDimension measureDimension = new CrosstabDimension(DIMENSION_MEASURES);
measureDimension.addCategory(MEASURE_ROW_COUNT);
measureDimension.addCategory(MEASURE_NULL_COUNT);
measureDimension.addCategory(MEASURE_BLANK_COUNT);
measureDimension.addCategory(MEASURE_ENTIRELY_UPPERCASE_COUNT);
measureDimension.addCategory(MEASURE_ENTIRELY_LOWERCASE_COUNT);
measureDimension.addCategory(MEASURE_TOTAL_CHAR_COUNT);
measureDimension.addCategory(MEASURE_MAX_CHARS);
measureDimension.addCategory(MEASURE_MIN_CHARS);
measureDimension.addCategory(MEASURE_AVG_CHARS);
measureDimension.addCategory(MEASURE_MAX_WHITE_SPACES);
measureDimension.addCategory(MEASURE_MIN_WHITE_SPACES);
measureDimension.addCategory(MEASURE_AVG_WHITE_SPACES);
measureDimension.addCategory(MEASURE_UPPERCASE_CHARS);
measureDimension.addCategory(MEASURE_UPPERCASE_CHARS_EXCL_FIRST_LETTERS);
measureDimension.addCategory(MEASURE_LOWERCASE_CHARS);
measureDimension.addCategory(MEASURE_DIGIT_CHARS);
measureDimension.addCategory(MEASURE_DIACRITIC_CHARS);
measureDimension.addCategory(MEASURE_NON_LETTER_CHARS);
measureDimension.addCategory(MEASURE_WORD_COUNT);
measureDimension.addCategory(MEASURE_MAX_WORDS);
measureDimension.addCategory(MEASURE_MIN_WORDS);
CrosstabDimension columnDimension = new CrosstabDimension(DIMENSION_COLUMN);
Crosstab<Number> crosstab =
new Crosstab<Number>(Number.class, columnDimension, measureDimension);
for (InputColumn<String> column : _columns) {
String columnName = column.getName();
StringAnalyzerColumnDelegate delegate = _columnDelegates.get(column);
columnDimension.addCategory(columnName);
final Integer numRows = delegate.getNumRows();
final Integer numNull = delegate.getNumNull();
final Integer numBlank = delegate.getNumBlank();
final Integer numEntirelyUppercase = delegate.getNumEntirelyUppercase();
final Integer numEntirelyLowercase = delegate.getNumEntirelyLowercase();
final Integer numChars = delegate.getNumChars();
final Integer maxChars = delegate.getMaxChars();
final Integer minChars = delegate.getMinChars();
final Integer numWords = delegate.getNumWords();
final Integer maxWords = delegate.getMaxWords();
final Integer minWords = delegate.getMinWords();
final Integer maxWhitespace = delegate.getMaxWhitespace();
final Integer minWhitespace = delegate.getMinWhitespace();
final Integer numUppercase = delegate.getNumUppercase();
final Integer numUppercaseExclFirstLetter = delegate.getNumUppercaseExclFirstLetter();
final Integer numLowercase = delegate.getNumLowercase();
final Integer numDigits = delegate.getNumDigit();
final Integer numDiacritics = delegate.getNumDiacritics();
final Integer numNonLetter = delegate.getNumNonLetter();
final AverageBuilder charAverageBuilder = delegate.getCharAverageBuilder();
final AverageBuilder blanksAverageBuilder = delegate.getWhitespaceAverageBuilder();
Double avgChars = null;
if (charAverageBuilder.getNumValues() > 0) {
avgChars = charAverageBuilder.getAverage();
}
Double avgBlanks = null;
if (blanksAverageBuilder.getNumValues() > 0) {
avgBlanks = blanksAverageBuilder.getAverage();
}
// begin entering numbers into the crosstab
CrosstabNavigator<Number> nav = crosstab.where(columnDimension, columnName);
nav.where(measureDimension, MEASURE_ROW_COUNT).put(numRows);
nav.where(measureDimension, MEASURE_NULL_COUNT).put(numNull);
if (numNull > 0) {
addAttachment(nav, delegate.getNullAnnotation(), column);
}
nav.where(measureDimension, MEASURE_BLANK_COUNT).put(numBlank);
if (numBlank > 0) {
addAttachment(nav, delegate.getBlankAnnotation(), column);
}
nav.where(measureDimension, MEASURE_ENTIRELY_UPPERCASE_COUNT).put(numEntirelyUppercase);
if (numEntirelyUppercase > 0) {
addAttachment(nav, delegate.getEntirelyUppercaseAnnotation(), column);
}
nav.where(measureDimension, MEASURE_ENTIRELY_LOWERCASE_COUNT).put(numEntirelyLowercase);
if (numEntirelyLowercase > 0) {
addAttachment(nav, delegate.getEntirelyLowercaseAnnotation(), column);
}
nav.where(measureDimension, MEASURE_TOTAL_CHAR_COUNT).put(numChars);
nav.where(measureDimension, MEASURE_MAX_CHARS).put(maxChars);
if (maxChars != null) {
addAttachment(nav, delegate.getMaxCharsAnnotation(), column);
}
nav.where(measureDimension, MEASURE_MIN_CHARS).put(minChars);
if (minChars != null) {
addAttachment(nav, delegate.getMinCharsAnnotation(), column);
}
nav.where(measureDimension, MEASURE_AVG_CHARS).put(avgChars);
nav.where(measureDimension, MEASURE_MAX_WHITE_SPACES).put(maxWhitespace);
if (maxWhitespace != null) {
addAttachment(nav, delegate.getMaxWhitespaceAnnotation(), column);
}
nav.where(measureDimension, MEASURE_MIN_WHITE_SPACES).put(minWhitespace);
if (minWhitespace != null) {
addAttachment(nav, delegate.getMinWhitespaceAnnotation(), column);
}
nav.where(measureDimension, MEASURE_AVG_WHITE_SPACES).put(avgBlanks);
nav.where(measureDimension, MEASURE_UPPERCASE_CHARS).put(numUppercase);
nav.where(measureDimension, MEASURE_UPPERCASE_CHARS_EXCL_FIRST_LETTERS)
.put(numUppercaseExclFirstLetter);
if (numUppercaseExclFirstLetter > 0) {
addAttachment(nav, delegate.getUppercaseExclFirstLetterAnnotation(), column);
}
nav.where(measureDimension, MEASURE_LOWERCASE_CHARS).put(numLowercase);
nav.where(measureDimension, MEASURE_DIGIT_CHARS).put(numDigits);
if (numDigits > 0) {
addAttachment(nav, delegate.getDigitAnnotation(), column);
}
nav.where(measureDimension, MEASURE_DIACRITIC_CHARS).put(numDiacritics);
if (numDiacritics > 0) {
addAttachment(nav, delegate.getDiacriticAnnotation(), column);
}
nav.where(measureDimension, MEASURE_NON_LETTER_CHARS).put(numNonLetter);
nav.where(measureDimension, MEASURE_WORD_COUNT).put(numWords);
nav.where(measureDimension, MEASURE_MAX_WORDS).put(maxWords);
if (maxWords != null) {
addAttachment(nav, delegate.getMaxWordsAnnotation(), column);
}
nav.where(measureDimension, MEASURE_MIN_WORDS).put(minWords);
if (minWords != null) {
addAttachment(nav, delegate.getMinWordsAnnotation(), column);
}
}
return new StringAnalyzerResult(_columns, crosstab);
}
