@Description("substring of given length starting at an index")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice substr(
@SqlType(VarcharType.class) Slice slice,
@SqlType(BigintType.class) long start,
@SqlType(BigintType.class) long length) {
if ((start == 0) || (length <= 0)) {
return Slices.EMPTY_SLICE;
}
if (start > 0) {
// make start zero-based
start--;
} else {
// negative start is relative to end of string
start += slice.length();
if (start < 0) {
return Slices.EMPTY_SLICE;
}
}
if ((start + length) > slice.length()) {
length = slice.length() - start;
}
if (start >= slice.length()) {
return Slices.EMPTY_SLICE;
}
return slice.slice((int) start, (int) length);
}
private static int firstNonSpace(Slice slice) {
for (int i = 0; i < slice.length(); i++) {
if (slice.getByte(i) != ' ') {
return i;
}
}
return slice.length();
}
@Description("converts all the alphabets in the string to upper case")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice upper(@SqlType(VarcharType.class) Slice slice) {
Slice upper = Slices.allocate(slice.length());
for (int i = 0; i < slice.length(); i++) {
upper.setByte(i, Ascii.toUpperCase((char) slice.getByte(i)));
}
return upper;
}
@Description("concatenates given strings")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice concat(
@SqlType(VarcharType.class) Slice str1, @SqlType(VarcharType.class) Slice str2) {
Slice concat = Slices.allocate(str1.length() + str2.length());
concat.setBytes(0, str1);
concat.setBytes(str1.length(), str2);
return concat;
}
@Description("reverses the given string")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice reverse(@SqlType(VarcharType.class) Slice slice) {
Slice reverse = Slices.allocate(slice.length());
for (int i = 0, j = slice.length() - 1; i < slice.length(); i++, j--) {
reverse.setByte(j, slice.getByte(i));
}
return reverse;
}
@Override
public void serialize(MaxOrMinByState state, BlockBuilder out) {
SliceOutput sliceOutput = new DynamicSliceOutput((int) state.getEstimatedSize());
int keyLength = 0;
if (state.getKey() != null && !state.getKey().isNull(0)) {
keyLength = state.getKey().getLength(0);
}
sliceOutput.writeInt(keyLength);
int valueLength = 0;
if (state.getValue() != null && !state.getValue().isNull(0)) {
valueLength = state.getValue().getLength(0);
}
sliceOutput.writeInt(valueLength);
if (state.getKey() != null && !state.getKey().isNull(0)) {
appendTo(keyType, sliceOutput, state.getKey());
}
if (state.getValue() != null && !state.getValue().isNull(0)) {
appendTo(valueType, sliceOutput, state.getValue());
}
Slice slice = sliceOutput.slice();
out.writeBytes(slice, 0, slice.length());
out.closeEntry();
}
@Description("removes spaces from the beginning of a string")
@ScalarFunction("ltrim")
@SqlType(VarcharType.class)
public static Slice leftTrim(@SqlType(VarcharType.class) Slice slice) {
int start = firstNonSpace(slice);
return slice.slice(start, slice.length() - start);
}
@Description("suffix starting at given index")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice substr(
@SqlType(VarcharType.class) Slice slice, @SqlType(BigintType.class) long start) {
return substr(slice, start, slice.length());
}
private static int lastNonSpace(Slice slice) {
for (int i = slice.length() - 1; i >= 0; i--) {
if (slice.getByte(i) != ' ') {
return i;
}
}
return -1;
}
private Block createZeroBlock(Type type, int rowsCount, Slice constantSlice) {
checkArgument(isSupportedType(type), "Unsupported type [%s]", type);
Slice slice;
// do not exceed varchar limit
if (isVarcharType(type)) {
slice =
constantSlice.slice(
0, Math.min(((VarcharType) type).getLength(), constantSlice.length()));
} else if (isLongDecimal(type)) {
slice = encodeScaledValue(ZERO);
} else {
slice = constantSlice;
}
BlockBuilder builder;
if (type instanceof FixedWidthType) {
builder = type.createBlockBuilder(new BlockBuilderStatus(), rowsCount);
} else {
builder = type.createBlockBuilder(new BlockBuilderStatus(), rowsCount, slice.length());
}
for (int i = 0; i < rowsCount; i++) {
Class<?> javaType = type.getJavaType();
if (javaType == boolean.class) {
type.writeBoolean(builder, false);
} else if (javaType == long.class) {
type.writeLong(builder, 0);
} else if (javaType == double.class) {
type.writeDouble(builder, 0.0);
} else if (javaType == Slice.class) {
requireNonNull(slice, "slice is null");
type.writeSlice(builder, slice, 0, slice.length());
} else {
throw new UnsupportedOperationException("Unknown javaType: " + javaType.getName());
}
}
return builder.build();
}
@Description("removes spaces from the beginning and end of a string")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice trim(@SqlType(VarcharType.class) Slice slice) {
int start = firstNonSpace(slice);
if (start == slice.length()) {
return Slices.EMPTY_SLICE;
}
int end = lastNonSpace(slice);
assert (end >= 0) && (end >= start);
return slice.slice(start, (end - start) + 1);
}
// TODO: Implement a more efficient string search
@Nullable
@Description("splits a string by a delimiter and returns the specified field (counting from one)")
@ScalarFunction
@SqlType(VarcharType.class)
public static Slice splitPart(
@SqlType(VarcharType.class) Slice string,
@SqlType(VarcharType.class) Slice delimiter,
@SqlType(BigintType.class) long index) {
checkArgument(index > 0, "Index must be greater than zero");
if (delimiter.length() == 0) {
if (index > string.length()) {
// index is too big, null is returned
return null;
}
return string.slice((int) (index - 1), 1);
}
int previousIndex = 0;
int matchCount = 0;
for (int i = 0; i <= (string.length() - delimiter.length()); i++) {
if (string.equals(i, delimiter.length(), delimiter, 0, delimiter.length())) {
matchCount++;
if (matchCount == index) {
return string.slice(previousIndex, i - previousIndex);
}
// noinspection AssignmentToForLoopParameter
i += (delimiter.length() - 1);
previousIndex = i + 1;
}
}
if (matchCount == index - 1) {
// returns last section of the split
return string.slice(previousIndex, string.length() - previousIndex);
}
// index is too big, null is returned
return null;
}
@Description("returns index of first occurrence of a substring (or 0 if not found)")
@ScalarFunction("strpos")
@SqlType(BigintType.class)
public static long stringPosition(
@SqlType(VarcharType.class) Slice string, @SqlType(VarcharType.class) Slice substring) {
if (substring.length() > string.length()) {
return 0;
}
for (int i = 0; i <= (string.length() - substring.length()); i++) {
if (string.equals(i, substring.length(), substring, 0, substring.length())) {
return i + 1;
}
}
return 0;
}
@Override
public BlockBuilder appendSlice(Slice value) {
return appendSlice(value, 0, value.length());
}
@SuppressWarnings("fallthrough")
public static DateTimeFormatter createDateTimeFormatter(Slice format) {
DateTimeFormatterBuilder builder = new DateTimeFormatterBuilder();
String formatString = format.toString(Charsets.UTF_8);
boolean escaped = false;
for (int i = 0; i < format.length(); i++) {
char character = formatString.charAt(i);
if (escaped) {
switch (character) {
case 'a': // %a Abbreviated weekday name (Sun..Sat)
builder.appendDayOfWeekShortText();
break;
case 'b': // %b Abbreviated month name (Jan..Dec)
builder.appendMonthOfYearShortText();
break;
case 'c': // %c Month, numeric (0..12)
builder.appendMonthOfYear(1);
break;
case 'd': // %d Day of the month, numeric (00..31)
builder.appendDayOfMonth(2);
break;
case 'e': // %e Day of the month, numeric (0..31)
builder.appendDayOfMonth(1);
break;
case 'f': // %f Microseconds (000000..999999)
builder.appendMillisOfSecond(6);
break;
case 'H': // %H Hour (00..23)
builder.appendHourOfDay(2);
break;
case 'h': // %h Hour (01..12)
case 'I': // %I Hour (01..12)
builder.appendClockhourOfHalfday(2);
break;
case 'i': // %i Minutes, numeric (00..59)
builder.appendMinuteOfHour(2);
break;
case 'j': // %j Day of year (001..366)
builder.appendDayOfYear(3);
break;
case 'k': // %k Hour (0..23)
builder.appendClockhourOfDay(1);
break;
case 'l': // %l Hour (1..12)
builder.appendClockhourOfHalfday(1);
break;
case 'M': // %M Month name (January..December)
builder.appendMonthOfYearText();
break;
case 'm': // %m Month, numeric (00..12)
builder.appendMonthOfYear(2);
break;
case 'p': // %p AM or PM
builder.appendHalfdayOfDayText();
break;
case 'r': // %r Time, 12-hour (hh:mm:ss followed by AM or PM)
builder
.appendClockhourOfHalfday(2)
.appendLiteral(':')
.appendMinuteOfHour(2)
.appendLiteral(':')
.appendSecondOfMinute(2)
.appendLiteral(' ')
.appendHalfdayOfDayText();
break;
case 'S': // %S Seconds (00..59)
case 's': // %s Seconds (00..59)
builder.appendSecondOfMinute(2);
break;
case 'T': // %T Time, 24-hour (hh:mm:ss)
builder
.appendHourOfDay(2)
.appendLiteral(':')
.appendMinuteOfHour(2)
.appendLiteral(':')
.appendSecondOfMinute(2);
break;
case 'v': // %v Week (01..53), where Monday is the first day of the week; used with %x
builder.appendWeekOfWeekyear(2);
break;
case 'x': // %x Year for the week, where Monday is the first day of the week, numeric,
// four digits; used with %v
builder.appendWeekyear(4, 4);
break;
case 'W': // %W Weekday name (Sunday..Saturday)
builder.appendDayOfWeekText();
break;
case 'w': // %w Day of the week (0=Sunday..6=Saturday)
builder.appendDayOfWeek(1);
break;
case 'Y': // %Y Year, numeric, four digits
builder.appendYear(4, 4);
break;
case 'y': // %y Year, numeric (two digits)
builder.appendYearOfCentury(2, 2);
break;
case 'U': // %U Week (00..53), where Sunday is the first day of the week
case 'u': // %u Week (00..53), where Monday is the first day of the week
case 'V': // %V Week (01..53), where Sunday is the first day of the week; used with %X
case 'X': // %X Year for the week where Sunday is the first day of the week, numeric, four
// digits; used with %V
case 'D': // %D Day of the month with English suffix (0th, 1st, 2nd, 3rd, …)
throw new PrestoException(
INVALID_FUNCTION_ARGUMENT.toErrorCode(),
String.format("%%%s not supported in date format string", character));
case '%': // %% A literal “%” character
builder.appendLiteral('%');
break;
default: // %<x> The literal character represented by <x>
builder.appendLiteral(character);
break;
}
escaped = false;
} else if (character == '%') {
escaped = true;
} else {
builder.appendLiteral(character);
}
}
try {
return builder.toFormatter();
} catch (UnsupportedOperationException e) {
throw new PrestoException(INVALID_FUNCTION_ARGUMENT.toErrorCode(), e);
}
}
public long getMemorySize() {
return slice.length() + sizeOf(positionOffsets.elements());
}
@Description("length of the given string")
@ScalarFunction
@SqlType(BigintType.class)
public static long length(@SqlType(VarcharType.class) Slice slice) {
return slice.length();
}