it.unimi.di.mg4j.tool
Class Paste

java.lang.Object
  extended by it.unimi.di.mg4j.tool.Combine
      extended by it.unimi.di.mg4j.tool.Paste

public final class Paste
extends Combine

Pastes several indices.

Pasting is a very slow way of combining indices: we assume that not only documents, but also document occurrences might be scattered throughout several indices. When a document appears in several indices, its occurrences in a given index are combined. We have two possibilities:

Standard pasting is used, for instance, to paste the batches of a virtual field generated by Scan; the latter takes care of numbering positions correctly. If, however, you index parts of the same document collection on different machines using the same VirtualDocumentResolver, then the resulting indices for virtual fields will have all position starting from zero, and they will need an incremental pasting to be combined correctly.

Conceptually, this operation is equivalent to splitting a collection vertically: each document is divided into a fixed number n of consecutive segments (possibly of length 0), and a set of n indices is created using the k-th segment of all documents. Pasting the resulting indices will produce an index that is identical to the index generated by the original collection. The behaviour is analogous to that of the UN*X paste command if documents are single-line lists of words.

Note that in case every document appears at most in one index pasting is equivalent to merging. It is, however, significantly slower, as the presence of the same document in several lists makes it necessary to scan completely the inverted lists to be pasted to compute the frequency. To do so, an in-memory buffer is allocated. If an inverted list does not fit in the memory buffer, it is spilled on disk. Sizing correctly the buffer, and choosing a fast file system for the temporary directory can significantly affect performance.

Warning: incremental pasting is very memory-intensive, as a list of sizes must be loaded for each index. You can use the URI option succinctsizes=1 to load sizes in a succinct format, which will ease the problem.

Since:
1.0
Author:
Sebastiano Vigna

Nested Class Summary
 
Nested classes/interfaces inherited from class it.unimi.di.mg4j.tool.Combine
Combine.GammaCodedIntIterator, Combine.IndexType
 
Field Summary
static int DEFAULT_MEMORY_BUFFER_SIZE
          The default size of the temporary bit stream buffer used while pasting.
protected  int[] doc
          The reference array of the document queue.
protected  IntHeapPriorityQueue documentQueue
          The queue containing document pointers (for remapped indices).
 
Fields inherited from class it.unimi.di.mg4j.tool.Combine
additionalProperties, bufferSize, DEFAULT_BUFFER_SIZE, frequency, hasCounts, hasPayloads, hasPositions, haveSumsMaxPos, index, indexIterator, indexReader, indexWriter, inputBasename, ioFactory, maxCount, metadataOnly, needsSizes, numberOfDocuments, numberOfOccurrences, numIndices, outputBasename, p, positionArray, predictedLengthNumBits, predictedSize, quasiSuccinctIndexWriter, size, sumsMaxPos, termQueue, usedIndex, variableQuantumIndexWriter
 
Constructor Summary
Paste(IOFactory ioFactory, String outputBasename, String[] inputBasename, boolean metadataOnly, boolean incremental, int bufferSize, File tempFileDir, int tempBufferSize, Map<CompressionFlags.Component,CompressionFlags.Coding> writerFlags, Combine.IndexType indexType, boolean skips, int quantum, int height, int skipBufferOrCacheSize, long logInterval)
          Pastes several indices into one.
Paste(IOFactory ioFactory, String outputBasename, String[] inputBasename, IntList delete, boolean metadataOnly, boolean incremental, int bufferSize, File tempFileDir, int tempBufferSize, Map<CompressionFlags.Component,CompressionFlags.Coding> writerFlags, Combine.IndexType indexType, boolean skips, int quantum, int height, int skipBufferOrCacheSize, long logInterval)
          Pastes several indices into one.
 
Method Summary
protected  int combine(int numUsedIndices, long occurrency)
          Combines several indices.
protected  int combineNumberOfDocuments()
          Combines the number of documents.
protected  int combineSizes(OutputBitStream sizesOutputBitStream)
          Combines size lists.
static void main(String[] arg)
           
 void run()
           
 
Methods inherited from class it.unimi.di.mg4j.tool.Combine
main, sizes
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

DEFAULT_MEMORY_BUFFER_SIZE

public static final int DEFAULT_MEMORY_BUFFER_SIZE
The default size of the temporary bit stream buffer used while pasting. Posting lists larger than this size will be precomputed on disk and then added to the index.

See Also:
Constant Field Values

doc

protected final int[] doc
The reference array of the document queue.


documentQueue

protected final IntHeapPriorityQueue documentQueue
The queue containing document pointers (for remapped indices).

Constructor Detail

Paste

public Paste(IOFactory ioFactory,
             String outputBasename,
             String[] inputBasename,
             boolean metadataOnly,
             boolean incremental,
             int bufferSize,
             File tempFileDir,
             int tempBufferSize,
             Map<CompressionFlags.Component,CompressionFlags.Coding> writerFlags,
             Combine.IndexType indexType,
             boolean skips,
             int quantum,
             int height,
             int skipBufferOrCacheSize,
             long logInterval)
      throws IOException,
             ConfigurationException,
             URISyntaxException,
             ClassNotFoundException,
             SecurityException,
             InstantiationException,
             IllegalAccessException,
             InvocationTargetException,
             NoSuchMethodException
Pastes several indices into one.

Parameters:
ioFactory - the factory that will be used to perform I/O.
outputBasename - the basename of the combined index.
inputBasename - the basenames of the input indices.
metadataOnly - if true, we save only metadata (term list, frequencies, global counts).
incremental - if true, we perform an incremental paste (needs sizes).
bufferSize - the buffer size for index readers.
tempFileDir - the directory of the temporary file used when pasting.
tempBufferSize - the size of the in-memory buffer used when pasting.
writerFlags - the flags for the index writer.
indexType - the type of the index to build.
skips - whether to insert skips in case interleaved is true.
quantum - the quantum of skipping structures; if negative, a percentage of space for variable-quantum indices (irrelevant if skips is false).
height - the height of skipping towers (irrelevant if skips is false).
skipBufferOrCacheSize - the size of the buffer used to hold temporarily inverted lists during the skipping structure construction, or the size of the bit cache used when building a quasi-succinct index.
logInterval - how often we log.
Throws:
IOException
ConfigurationException
URISyntaxException
ClassNotFoundException
SecurityException
InstantiationException
IllegalAccessException
InvocationTargetException
NoSuchMethodException

Paste

public Paste(IOFactory ioFactory,
             String outputBasename,
             String[] inputBasename,
             IntList delete,
             boolean metadataOnly,
             boolean incremental,
             int bufferSize,
             File tempFileDir,
             int tempBufferSize,
             Map<CompressionFlags.Component,CompressionFlags.Coding> writerFlags,
             Combine.IndexType indexType,
             boolean skips,
             int quantum,
             int height,
             int skipBufferOrCacheSize,
             long logInterval)
      throws IOException,
             ConfigurationException,
             URISyntaxException,
             ClassNotFoundException,
             SecurityException,
             InstantiationException,
             IllegalAccessException,
             InvocationTargetException,
             NoSuchMethodException
Pastes several indices into one.

Parameters:
ioFactory - the factory that will be used to perform I/O.
outputBasename - the basename of the combined index.
inputBasename - the basenames of the input indices.
delete - a monotonically increasing list of integers representing documents that will be deleted from the output index, or null.
metadataOnly - if true, we save only metadata (term list, frequencies, global counts).
incremental - if true, we perform an incremental paste (needs sizes).
bufferSize - the buffer size for index readers.
tempFileDir - the directory of the temporary file used when pasting.
tempBufferSize - the size of the in-memory buffer used when pasting.
writerFlags - the flags for the index writer.
indexType - the type of the index to build.
skips - whether to insert skips in case interleaved is true.
quantum - the quantum of skipping structures; if negative, a percentage of space for variable-quantum indices (irrelevant if skips is false).
height - the height of skipping towers (irrelevant if skips is false).
skipBufferOrCacheSize - the size of the buffer used to hold temporarily inverted lists during the skipping structure construction, or the size of the bit cache used when building a quasi-succinct index.
logInterval - how often we log.
Throws:
IOException
ConfigurationException
URISyntaxException
ClassNotFoundException
SecurityException
InstantiationException
IllegalAccessException
InvocationTargetException
NoSuchMethodException
Method Detail

combineNumberOfDocuments

protected int combineNumberOfDocuments()
Description copied from class: Combine
Combines the number of documents.

Specified by:
combineNumberOfDocuments in class Combine
Returns:
the number of documents of the combined index.

combineSizes

protected int combineSizes(OutputBitStream sizesOutputBitStream)
                    throws IOException
Description copied from class: Combine
Combines size lists.

Specified by:
combineSizes in class Combine
Returns:
the maximum size of a document in the combined index.
Throws:
IOException

combine

protected int combine(int numUsedIndices,
                      long occurrency)
               throws IOException
Description copied from class: Combine
Combines several indices.

When this method is called, exactly numUsedIndices entries of Combine.usedIndex contain, in increasing order, the indices containing inverted lists for the current term. Implementations of this method must combine the inverted list and return the total frequency.

Specified by:
combine in class Combine
Parameters:
numUsedIndices - the number of valid entries in Combine.usedIndex.
occurrency - the occurrency of the term (used only when building Combine.IndexType.QUASI_SUCCINCT indices).
Returns:
the total frequency.
Throws:
IOException

run

public void run()
         throws ConfigurationException,
                IOException
Overrides:
run in class Combine
Throws:
ConfigurationException
IOException

main

public static void main(String[] arg)
                 throws ConfigurationException,
                        SecurityException,
                        com.martiansoftware.jsap.JSAPException,
                        IOException,
                        URISyntaxException,
                        ClassNotFoundException,
                        InstantiationException,
                        IllegalAccessException,
                        InvocationTargetException,
                        NoSuchMethodException,
                        IllegalArgumentException
Throws:
ConfigurationException
SecurityException
com.martiansoftware.jsap.JSAPException
IOException
URISyntaxException
ClassNotFoundException
InstantiationException
IllegalAccessException
InvocationTargetException
NoSuchMethodException
IllegalArgumentException