use of io.trino.spi.function.SqlType in project trino by trinodb.
the class StringFunctions method hammingDistance.
@Description("Computes Hamming distance between two strings")
@ScalarFunction
@LiteralParameters({ "x", "y" })
@SqlType(StandardTypes.BIGINT)
public static long hammingDistance(@SqlType("varchar(x)") Slice left, @SqlType("varchar(y)") Slice right) {
int distance = 0;
int leftPosition = 0;
int rightPosition = 0;
while (leftPosition < left.length() && rightPosition < right.length()) {
int codePointLeft = tryGetCodePointAt(left, leftPosition);
int codePointRight = tryGetCodePointAt(right, rightPosition);
// the following code treats them as equal if they happen to be of the same length
if (codePointLeft != codePointRight) {
distance++;
}
leftPosition += codePointLeft > 0 ? lengthOfCodePoint(codePointLeft) : -codePointLeft;
rightPosition += codePointRight > 0 ? lengthOfCodePoint(codePointRight) : -codePointRight;
}
checkCondition(leftPosition == left.length() && rightPosition == right.length(), INVALID_FUNCTION_ARGUMENT, "The input strings to hamming_distance function must have the same length");
return distance;
}
use of io.trino.spi.function.SqlType in project trino by trinodb.
the class StringFunctions method fromUtf8.
@Description("Decodes the UTF-8 encoded string")
@ScalarFunction
@LiteralParameters("x")
@SqlType(StandardTypes.VARCHAR)
public static Slice fromUtf8(@SqlType(StandardTypes.VARBINARY) Slice slice, @SqlType("varchar(x)") Slice replacementCharacter) {
int count = countCodePoints(replacementCharacter);
if (count > 1) {
throw new TrinoException(INVALID_FUNCTION_ARGUMENT, "Replacement character string must empty or a single character");
}
OptionalInt replacementCodePoint;
if (count == 1) {
try {
replacementCodePoint = OptionalInt.of(getCodePointAt(replacementCharacter, 0));
} catch (InvalidUtf8Exception e) {
throw new TrinoException(INVALID_FUNCTION_ARGUMENT, "Invalid replacement character");
}
} else {
replacementCodePoint = OptionalInt.empty();
}
return SliceUtf8.fixInvalidUtf8(slice, replacementCodePoint);
}
use of io.trino.spi.function.SqlType in project trino by trinodb.
the class StringFunctions method replace.
@Description("Greedily replaces occurrences of a pattern with a string")
@ScalarFunction
@LiteralParameters({ "x", "y", "z", "u" })
@Constraint(variable = "u", expression = "min(2147483647, x + z * (x + 1))")
@SqlType("varchar(u)")
public static Slice replace(@SqlType("varchar(x)") Slice str, @SqlType("varchar(y)") Slice search, @SqlType("varchar(z)") Slice replace) {
// Empty search?
if (search.length() == 0) {
// With empty `search` we insert `replace` in front of every character and the end
Slice buffer = Slices.allocate((countCodePoints(str) + 1) * replace.length() + str.length());
// Always start with replace
buffer.setBytes(0, replace);
int indexBuffer = replace.length();
// After every code point insert `replace`
int index = 0;
while (index < str.length()) {
int codePointLength = lengthOfCodePointSafe(str, index);
// Append current code point
buffer.setBytes(indexBuffer, str, index, codePointLength);
indexBuffer += codePointLength;
// Append `replace`
buffer.setBytes(indexBuffer, replace);
indexBuffer += replace.length();
// Advance pointer to current code point
index += codePointLength;
}
return buffer;
}
// Allocate a reasonable buffer
Slice buffer = Slices.allocate(str.length());
int index = 0;
int indexBuffer = 0;
while (index < str.length()) {
int matchIndex = str.indexOf(search, index);
// Found a match?
if (matchIndex < 0) {
// No match found so copy the rest of string
int bytesToCopy = str.length() - index;
buffer = Slices.ensureSize(buffer, indexBuffer + bytesToCopy);
buffer.setBytes(indexBuffer, str, index, bytesToCopy);
indexBuffer += bytesToCopy;
break;
}
int bytesToCopy = matchIndex - index;
buffer = Slices.ensureSize(buffer, indexBuffer + bytesToCopy + replace.length());
// Non empty match?
if (bytesToCopy > 0) {
buffer.setBytes(indexBuffer, str, index, bytesToCopy);
indexBuffer += bytesToCopy;
}
// Non empty replace?
if (replace.length() > 0) {
buffer.setBytes(indexBuffer, replace);
indexBuffer += replace.length();
}
// Continue searching after match
index = matchIndex + search.length();
}
return buffer.slice(0, indexBuffer);
}
use of io.trino.spi.function.SqlType in project trino by trinodb.
the class RepeatFunction method repeat.
@TypeParameter("T")
@SqlType("array(T)")
public static Block repeat(@TypeParameter("T") Type type, @SqlNullable @SqlType("T") Object element, @SqlType(StandardTypes.INTEGER) long count) {
BlockBuilder blockBuilder = createBlockBuilder(type, count);
if (element == null) {
return repeatNullValues(blockBuilder, count);
}
if (count > 0) {
type.writeObject(blockBuilder, element);
checkMaxSize(blockBuilder.getSizeInBytes(), count);
}
for (int i = 1; i < count; i++) {
type.writeObject(blockBuilder, element);
}
return blockBuilder.build();
}
use of io.trino.spi.function.SqlType in project trino by trinodb.
the class UrlFunctions method urlExtractParameter.
@SqlNullable
@Description("Extract query parameter from url")
@ScalarFunction
@LiteralParameters({ "x", "y" })
@SqlType("varchar(x)")
public static Slice urlExtractParameter(@SqlType("varchar(x)") Slice url, @SqlType("varchar(y)") Slice parameterName) {
URI uri = parseUrl(url);
if ((uri == null) || (uri.getRawQuery() == null)) {
return null;
}
String parameter = parameterName.toStringUtf8();
Iterable<String> queryArgs = QUERY_SPLITTER.split(uri.getRawQuery());
for (String queryArg : queryArgs) {
Iterator<String> arg = ARG_SPLITTER.split(queryArg).iterator();
if (arg.next().equals(parameter)) {
if (arg.hasNext()) {
return decodeUrl(arg.next());
}
// first matched key is empty
return Slices.EMPTY_SLICE;
}
}
// no key matched
return null;
}
Aggregations