org.apache.lucene.util.fst

Class Util

java.lang.Object

org.apache.lucene.util.fst.Util

public final class Util
extends Object

Static helper methods.

WARNING: This API is experimental and might change in incompatible ways in the next release.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Util.MinResult<T> Holds a single input (IntsRef) + output, returned by shortestPaths(org.apache.lucene.util.fst.FST<T>, org.apache.lucene.util.fst.FST.Arc<T>, java.util.Comparator<T>, int).

Method Summary

Methods

Modifier and Type	Method and Description
static <T> T	get(FST<T> fst, BytesRef input) Looks up the output for this input, or null if the input is not accepted
static <T> T	get(FST<T> fst, IntsRef input) Looks up the output for this input, or null if the input is not accepted.
static IntsRef	getByOutput(FST<Long> fst, long targetOutput) Reverse lookup (lookup by output instead of by input), in the special case when your FSTs outputs are strictly ascending.
static <T> Util.MinResult<T>[]	shortestPaths(FST<T> fst, FST.Arc<T> fromNode, Comparator<T> comparator, int topN) Starting from node, find the top N min cost completions to a final node.
static BytesRef	toBytesRef(IntsRef input, BytesRef scratch) Just converts IntsRef to BytesRef; you must ensure the int values fit into a byte.
static <T> void	toDot(FST<T> fst, Writer out, boolean sameRank, boolean labelStates) Dumps an FST to a GraphViz's dot language description for visualization.
static IntsRef	toIntsRef(BytesRef input, IntsRef scratch) Just takes unsigned byte values from the BytesRef and converts into an IntsRef.
static IntsRef	toUTF32(char[] s, int offset, int length, IntsRef scratch) Decodes the Unicode codepoints from the provided char[] and places them in the provided scratch IntsRef, which must not be null, returning it.
static IntsRef	toUTF32(CharSequence s, IntsRef scratch) Decodes the Unicode codepoints from the provided CharSequence and places them in the provided scratch IntsRef, which must not be null, returning it.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

get

public static <T> T get(FST<T> fst,
        IntsRef input)
             throws IOException

Looks up the output for this input, or null if the input is not accepted.

Throws:

IOException

get

public static <T> T get(FST<T> fst,
        BytesRef input)
             throws IOException

Looks up the output for this input, or null if the input is not accepted

Throws:

IOException

getByOutput

public static IntsRef getByOutput(FST<Long> fst,
                  long targetOutput)
                           throws IOException

Reverse lookup (lookup by output instead of by input), in the special case when your FSTs outputs are strictly ascending. This locates the input/output pair where the output is equal to the target, and will return null if that output does not exist.

NOTE: this only works with FST, only works when the outputs are ascending in order with the inputs and only works when you shared the outputs (pass doShare=true to PositiveIntOutputs.getSingleton(boolean)). For example, simple ordinals (0, 1, 2, ...), or file offets (when appending to a file) fit this.

Throws:

IOException

shortestPaths

public static <T> Util.MinResult<T>[] shortestPaths(FST<T> fst,
                                    FST.Arc<T> fromNode,
                                    Comparator<T> comparator,
                                    int topN)
                                         throws IOException

Starting from node, find the top N min cost completions to a final node.

NOTE: you must share the outputs when you build the FST (pass doShare=true to PositiveIntOutputs.getSingleton(boolean)).

Throws:

IOException

toDot

public static <T> void toDot(FST<T> fst,
             Writer out,
             boolean sameRank,
             boolean labelStates)
                  throws IOException

Dumps an FST to a GraphViz's dot language description for visualization. Example of use:

 PrintWriter pw = new PrintWriter("out.dot");
 Util.toDot(fst, pw, true, true);
 pw.close();
 

and then, from command line:

 dot -Tpng -o out.png out.dot
 

Note: larger FSTs (a few thousand nodes) won't even render, don't bother.

Parameters:

sameRank - If true, the resulting dot file will try to order states in layers of breadth-first traversal. This may mess up arcs, but makes the output FST's structure a bit clearer.

labelStates - If true states will have labels equal to their offsets in their binary format. Expands the graph considerably.

Throws:

IOException

See Also:

"http://www.graphviz.org/"

toUTF32

public static IntsRef toUTF32(CharSequence s,
              IntsRef scratch)

Decodes the Unicode codepoints from the provided CharSequence and places them in the provided scratch IntsRef, which must not be null, returning it.

toUTF32

public static IntsRef toUTF32(char[] s,
              int offset,
              int length,
              IntsRef scratch)

Decodes the Unicode codepoints from the provided char[] and places them in the provided scratch IntsRef, which must not be null, returning it.

toIntsRef

public static IntsRef toIntsRef(BytesRef input,
                IntsRef scratch)

Just takes unsigned byte values from the BytesRef and converts into an IntsRef.

toBytesRef

public static BytesRef toBytesRef(IntsRef input,
                  BytesRef scratch)

Just converts IntsRef to BytesRef; you must ensure the int values fit into a byte.