LUCENE-3892: fix maxChain bug; support numBits=0 (all ints == 0)

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/branches/pforcodec_3892@1360993 13f79535-47bb-0310-9956-ffa450edef68

lucene/core/src/java/org/apache/lucene/codecs/pfor/ForPostingsFormat.java

@@ -17,28 +17,27 @@ package org.apache.lucene.codecs.pfor;
  * limitations under the License.
  */
 
-import java.io.IOException;
 import java.util.Set;
+import java.io.IOException;
 
-import org.apache.lucene.codecs.BlockTreeTermsReader;
+import org.apache.lucene.util.IOUtils;
-import org.apache.lucene.codecs.BlockTreeTermsWriter;
+import org.apache.lucene.store.IOContext;
+import org.apache.lucene.index.SegmentInfo;
+import org.apache.lucene.index.SegmentWriteState;
+import org.apache.lucene.index.SegmentReadState;
 import org.apache.lucene.codecs.FieldsConsumer;
 import org.apache.lucene.codecs.FieldsProducer;
+import org.apache.lucene.codecs.BlockTreeTermsWriter;
+import org.apache.lucene.codecs.BlockTreeTermsReader;
+import org.apache.lucene.codecs.TermsIndexReaderBase;
+import org.apache.lucene.codecs.TermsIndexWriterBase;
 import org.apache.lucene.codecs.FixedGapTermsIndexReader;
 import org.apache.lucene.codecs.FixedGapTermsIndexWriter;
 import org.apache.lucene.codecs.PostingsFormat;
-import org.apache.lucene.codecs.PostingsReaderBase;
 import org.apache.lucene.codecs.PostingsWriterBase;
-import org.apache.lucene.codecs.TermsIndexReaderBase;
+import org.apache.lucene.codecs.PostingsReaderBase;
-import org.apache.lucene.codecs.TermsIndexWriterBase;
 import org.apache.lucene.codecs.sep.SepPostingsReader;
 import org.apache.lucene.codecs.sep.SepPostingsWriter;
-import org.apache.lucene.index.SegmentInfo;
-import org.apache.lucene.index.SegmentReadState;
-import org.apache.lucene.index.SegmentWriteState;
-import org.apache.lucene.store.IOContext;
-import org.apache.lucene.util.BytesRef;
-import org.apache.lucene.util.IOUtils;
 
 /**
  * Pass ForFactory to a PostingsWriter/ReaderBase, and get 

lucene/core/src/java/org/apache/lucene/codecs/pfor/ForUtil.java

@@ -63,7 +63,7 @@ public class ForUtil {
 
     int header = intBuffer.get();
     int numInts = (header & MASK[8]) + 1;
-    int numBits = ((header >> 8) & MASK[5]) + 1;
+    int numBits = ((header >> 8) & MASK[6]);
 
     decompressCore(intBuffer, data, numBits);
 
@@ -77,10 +77,11 @@ public class ForUtil {
    */
   static void decompressCore(IntBuffer intBuffer, int[] data, int numBits) {
     assert numBits<=32;
-    assert numBits>=1;
+    assert numBits>=0;
 
-    // TODO: PackedIntsDecompress is hardewired to size==129 only
+    // TODO: PackedIntsDecompress is hardewired to size==128 only
     switch(numBits) {
+      case 0: PackedIntsDecompress.decode0(intBuffer, data); break;
       case 1: PackedIntsDecompress.decode1(intBuffer, data); break;
       case 2: PackedIntsDecompress.decode2(intBuffer, data); break;
       case 3: PackedIntsDecompress.decode3(intBuffer, data); break;
@@ -145,7 +146,7 @@ public class ForUtil {
    * Estimate best num of frame bits according to the largest value.
    */
   static int getNumBits(final int[] data, int size) {
-    int optBits=1;
+    int optBits=0;
     for (int i=0; i<size; ++i) {
       while ((data[i] & ~MASK[optBits]) != 0) {
         optBits++;
@@ -158,12 +159,12 @@ public class ForUtil {
    * Generate the 4 byte header, which contains (from lsb to msb):
    *
    * - 8 bits for uncompressed int num - 1 (use up to 7 bits i.e 128 actually)
-   * - 5 bits for num of frame bits - 1
+   * - 6 bits for num of frame bits
    * - other bits unused
    *
    */
   static int getHeader(int numInts, int numBits) {
     return  (numInts-1)
-          | ((numBits-1) << 8);
+          | ((numBits) << 8);
   }
 }

lucene/core/src/java/org/apache/lucene/codecs/pfor/PForUtil.java

@@ -53,7 +53,9 @@ public final class PForUtil extends ForUtil {
 
     // the max value possible for current exception pointer, 
     // value of the first pointer is limited by header as 254
-    long maxChain = (1<<8) - 2;  
+    // (first exception ranges from -1 ~ 254)
+    long maxChainFirst = 254;
+    long maxChain = maxChainFirst + 1;  
 
     boolean conValue, conForce, conEnd;
     int i=0;
@@ -77,41 +79,35 @@ public final class PForUtil extends ForUtil {
     }
 
     /** encode normal area, record exception positions */
-    i=0;
     excNum = 0;
     if (excFirstPos < 0) { // no exception 
-      for (; i<size; ++i) {
+      for (i=0; i<size; ++i) {
         encodeNormalValue(intBuffer,i,data[i], numBits);
       }
       excLastPos = -1;
     } else {
-      for (; i<excFirstPos; ++i) {
+      for (i=0; i<excFirstPos; ++i) {
         encodeNormalValue(intBuffer,i,data[i], numBits);
       }
       maxChain = 1L<<numBits;
-      excLastPos = -1;
+      excLastPos = excFirstPos;
-      for (; i<size; ++i) {
+      excNum = i<size? 1:0;
+      for (i=excFirstPos+1; i<size; ++i) {
         conValue = ((data[i] & MASK[numBits]) != data[i]); // value exception
         conForce = (i >= maxChain + excLastPos);           // force exception
         conEnd = (excNum == excNumBase);                   // following forced ignored
         if ((!conValue && !conForce) || conEnd) {
           encodeNormalValue(intBuffer,i,data[i], numBits);
         } else {
-          if (excLastPos >= 0) {
+          encodeNormalValue(intBuffer, excLastPos, i-excLastPos-1, numBits); 
-            encodeNormalValue(intBuffer, excLastPos, i-excLastPos-1, numBits); 
-          }
           excNum++;
           excLastPos = i;
         }
       }
-      if (excLastPos >= 0) { 
-        encodeNormalValue(intBuffer, excLastPos, (i-excLastPos-1)&MASK[numBits], numBits); // mask out suppressed force exception
-      }
     }
   
     /** encode exception area */
-    i=0;
+    for (i=0; i<excNum; ++i) {
-    for (; i<excNum; ++i) {
       if (excBytes < 2 && (excValues[i] & ~MASK[8]) != 0) {
         excBytes=2;
       }
@@ -139,11 +135,11 @@ public final class PForUtil extends ForUtil {
     intBuffer.rewind();
 
     int header = intBuffer.get();
-    int numInts = (header & MASK[8]) + 1;
+    int numInts = (header & MASK[8]);
     int excNum = ((header >> 8) & MASK[8]) + 1;
     int excFirstPos = ((header >> 16) & MASK[8]) - 1;
-    int excBytes = PER_EXCEPTION_SIZE[(header >> 29) & MASK[2]];
+    int excBytes = PER_EXCEPTION_SIZE[(header >> 30) & MASK[2]];
-    int numBits = ((header >> 24) & MASK[5]) + 1;
+    int numBits = ((header >> 24) & MASK[6]);
 
     decompressCore(intBuffer, data, numBits);
 
@@ -157,6 +153,11 @@ public final class PForUtil extends ForUtil {
     intBuffer.put(0, header);
   }
 
+  /**
+   * Encode exception values into exception area.
+   * The width for each exception will be fixed as:
+   * 1, 2, or 4 byte(s).
+   */
   static void encodeExcValues(IntBuffer intBuffer, int[] values, int num, int perbytes, int byteOffset) {
     if (num == 0)
       return;
@@ -251,8 +252,10 @@ public final class PForUtil extends ForUtil {
    * It will run 32 times.
    */
   static int getNumBits(final int[] data, int size) {
+    if (isAllZero(data))
+      return 0;
     int optBits=1;
-    int optSize=estimateCompressedSize(data,size,1);
+    int optSize=estimateCompressedSize(data,size,optBits);
     for (int i=2; i<=32; ++i) {
       int curSize=estimateCompressedSize(data,size,i);
       if (curSize<optSize) {
@@ -263,6 +266,16 @@ public final class PForUtil extends ForUtil {
     return optBits;
   }
 
+  static boolean isAllZero(final int[] data) {
+    int len=data.length;
+    for (int i=0; i<len; i++) {
+      if (data[i] != 0) {
+        return false;
+      }
+    }
+    return true;
+  }
+
   /**
    * Iterate the whole block to get maximum exception bits, 
    * and estimate compressed size without forced exception.
@@ -303,16 +316,40 @@ public final class PForUtil extends ForUtil {
    *
    * 8 bits for the index of the first exception + 1 (when no exception, this is 0)
    *
-   * 5 bits for num of frame bits - 1
+   * 6 bits for num of frame bits
    * 2 bits for the exception code: 00: byte, 01: short, 10: int
-   * 1 bit unused
    *
    */
+  // TODO: exception num should never be equal with uncompressed int num!!!
+  // first exception ranges from -1 ~ 255
+  // the problem is that we don't need first exception to be -1 ...
+  // it is ok to range from 0~255, and judge exception for exception num (0~255)
+  // uncompressed int num: (1~256)
   static int getHeader(int numInts, int numBits, int excNum, int excFirstPos, int excBytes) {
     return  (numInts-1)
           | (((excNum-1) & MASK[8]) << 8)
           | ((excFirstPos+1) << 16)
-          | ((numBits-1) << 24)
+          | ((numBits) << 24)
-          | ((excBytes/2) << 29);
+          | ((excBytes/2) << 30);
+  }
+
+
+  /** 
+   * Expert: get metadata from header. 
+   */
+  public static int getNumInts(int header) {
+    return (header & MASK[8]) + 1;
+  }
+  public static int getExcNum(int header) {
+    return ((header >> 8) & MASK[8]) + 1;
+  }
+  public static int getFirstPos(int header) {
+    return ((header >> 16) & MASK[8]) - 1;
+  }
+  public static int getExcBytes(int header) {
+    return PER_EXCEPTION_SIZE[(header >> 30) & MASK[2]];
+  }
+  public static int getNumBits(int header) {
+    return ((header >> 24) & MASK[6]);
   }
 }

lucene/core/src/java/org/apache/lucene/codecs/pfor/PackedIntsDecompress.java

@@ -19,6 +19,7 @@ package org.apache.lucene.codecs.pfor;
 /* This code is generated, do not modify. See gendecompress.py */
 
 import java.nio.IntBuffer;
+import java.util.Arrays;
 
 final class PackedIntsDecompress {
 
@@ -26,6 +27,9 @@ final class PackedIntsDecompress {
 
   // NOTE: hardwired to blockSize == 128
 
+  public static void decode0(final IntBuffer compressedBuffer, final int[] output) {
+    Arrays.fill(output, 0);
+  }
   public static void decode1(final IntBuffer compressedBuffer, final int[] output) {
     final int numFrameBits = 1;
     final int mask = (int) ((1L<<numFrameBits) - 1);

lucene/core/src/java/org/apache/lucene/codecs/pfor/gendecompress.py

@@ -69,13 +69,18 @@ def genDecompress():
 
     w("\n/* This code is generated, do not modify. See gendecompress.py */\n\n")
 
-    w("import java.nio.IntBuffer;\n\n")
+    w("import java.nio.IntBuffer;\n")
+    w("import java.util.Arrays;\n\n")
 
     w("final class PackedIntsDecompress {\n")
 
     w('\n  // nocommit: assess perf of this to see if specializing is really needed\n')
     w('\n  // NOTE: hardwired to blockSize == 128\n\n')
 
+    w('  public static void decode0(final IntBuffer compressedBuffer, final int[] output) {\n')
+    w('    Arrays.fill(output, 0);\n')
+    w('  }\n')
+
     for numFrameBits in xrange(1, 33):
       w('  public static void decode%d(final IntBuffer compressedBuffer, final int[] output) {\n' % numFrameBits)
       w('    final int numFrameBits = %d;\n' % numFrameBits)

lucene/core/src/test/org/apache/lucene/codecs/pfor/TestPForUtil.java

@@ -28,7 +28,7 @@ import org.apache.lucene.util.LuceneTestCase;
 
 /**
  * Test the core utility for PFor compress and decompress
- * We don't provide test case for For encoder/decoder, since
+ * We don't specially provide test case for For encoder/decoder, since
  * PFor is a extended version of For, and most methods will be reused 
  * here.
  */
@@ -46,37 +46,157 @@ public class TestPForUtil extends LuceneTestCase {
   }
 
   /**
-   * Test correctness of ignored forced exception.
+   * Should not encode extra information other than header
-   * The trailing forced exceptions shouldn't be reverted
-   * since they're not necessary. 
    */
-  public void testForcedException() throws Exception {
+  public void testPForAllZeros() throws Exception {
+    int sz=ForPostingsFormat.DEFAULT_BLOCK_SIZE;
+    int ensz;
+    int[] data=new int[sz];
+    byte[] res = new byte[4+sz*8];
+    int[] copy = new int[sz];
+    IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
+
+    Arrays.fill(data,0);
+    ensz = ForUtil.compress(data,sz,resBuffer); // test For
+    ForUtil.decompress(resBuffer,copy);
+    assert ensz == 4;
+    assert cmp(data,sz,copy,sz)==true;
+
+    Arrays.fill(data,0);
+    ensz = PForUtil.compress(data,sz,resBuffer); // test PFor
+    PForUtil.decompress(resBuffer,copy);
+    assert ensz == 4;
+    assert cmp(data,sz,copy,sz)==true;
+  }
+
+  public void testForAllZeros() throws Exception {
+    int sz=ForPostingsFormat.DEFAULT_BLOCK_SIZE;
+    int ensz;
+    int[] data=new int[sz];
+    byte[] res = new byte[4+sz*8];
+    int[] copy = new int[sz];
+    IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
+
+    ensz = ForUtil.compress(data,sz,resBuffer);
+
+  }
+
+  /**
+   * Test correctness of forced exception.
+   * the forced ones should exactly fit max chain 
+   */
+  public void testForcedExceptionDistance() throws Exception {
     initRandom();
     int sz=ForPostingsFormat.DEFAULT_BLOCK_SIZE;
+    int[] data=new int[sz];
+    byte[] res = new byte[4+sz*8];
+    int[] copy = new int[sz];
+    IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
+    int numBits = gen.nextInt(5)+1;
+
+    int i,j;
+    int pace, ensz, header;
+    int expect, got;
+
+    // fill exception value with same pace, there should
+    // be no forced exceptions.
+    createDistribution(data, sz, 1, MASK[numBits], MASK[numBits]);
+    pace = 1<<numBits;
+    for (i=0,j=0; i<sz; i+=pace) {
+      int exc = gen.nextInt();
+      data[i] = (exc & 0xffff0000) == 0 ? exc | 0xffff0000 : exc;
+      j++;
+    }
+    ensz = PForUtil.compress(data,sz,resBuffer);
+    header = resBuffer.get(0);
+    expect = j; 
+    got = PForUtil.getExcNum(header);
+    assert expect == got: expect+" expected but got "+got;
+
+    // there should exactly one forced exception before each
+    // exception when i>0
+    createDistribution(data, sz, 1, MASK[numBits], MASK[numBits]);
+    pace = (1<<numBits)+1;
+    for (i=0,j=0; i<sz; i+=pace) {
+      int exc = gen.nextInt();
+      data[i] = (exc & 0xffff0000) == 0 ? exc | 0xffff0000 : exc;
+      j++;
+    }
+    ensz = PForUtil.compress(data,sz,resBuffer);
+    header = resBuffer.get(0);
+    expect = 2*(j-1)+1; 
+    got = PForUtil.getExcNum(header);
+    assert expect == got: expect+" expected but got "+got;
+
+
+    // two forced exception  
+    createDistribution(data, sz, 1, MASK[numBits], MASK[numBits]);
+    pace = (1<<numBits)*2+1;
+    for (i=0,j=0; i<sz; i+=pace) {
+      int exc = gen.nextInt();
+      data[i] = (exc & 0xffff0000) == 0 ? exc | 0xffff0000 : exc;
+      j++;
+    }
+    ensz = PForUtil.compress(data,sz,resBuffer);
+    header = resBuffer.get(0);
+    expect = 3*(j-1)+1; 
+    got = PForUtil.getExcNum(header);
+    assert expect == got: expect+" expected but got "+got;
+
+  }
+  /**
+   * Test correctness of ignored forced exception.
+   * The trailing forced exceptions should always be reverted
+   * since they're not necessary. 
+   */
+  public void testTrailingForcedException() throws Exception {
+    initRandom();
+    int sz=ForPostingsFormat.DEFAULT_BLOCK_SIZE;
+    assert sz % 32 == 0;
     Integer[] buff= new Integer[sz];
     int[] data = new int[sz];
     int[] copy = new int[sz];
     byte[] res = new byte[4+sz*8];
     IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
-    for (int i=0; i<sz-1; ++i)
-      buff[i]=gen.nextInt() & 1;
-    buff[sz-1]=gen.nextInt() & 0xffffffff;   // create only one exception
 
-    Collections.shuffle(Arrays.asList(buff),gen);
+    int excIndex = gen.nextInt(sz/2);
+    int excValue = gen.nextInt();
+    if ((excValue & 0xffff0000) == 0) {
+      excValue |= 0xffff0000; // always prepare a 4 bytes exception
+    }
+
+    // make value of numFrameBits to be small, 
+    // thus easy to get forced exceptions
+    for (int i=0; i<sz; ++i) {
+      buff[i]=gen.nextInt() & 1;
+    }
+    // create only one value exception
+    buff[excIndex]=excValue;
+
     for (int i=0; i<sz; ++i)
       data[i] = buff[i];
 
     int ensz = PForUtil.compress(data,sz,resBuffer);
 
-    assert (ensz <= sz*8+4);  // must not exceed the loose upperbound
+    assert (ensz <= sz*8+4): ensz+" > "+sz*8+4;  // must not exceed the loose upperbound
+    assert (ensz >= 8);       // at least we have a header along with an exception, right?
 
     resBuffer.rewind();
     PForUtil.decompress(resBuffer,copy);
 
-    //println(getHex(data,sz)+"\n");
+//    println(getHex(data,sz)+"\n");
-    //println(getHex(res,ensz)+"\n");
+//    println(getHex(res,ensz)+"\n");
-    //println(getHex(copy,sz)+"\n");
+//    println(getHex(copy,sz)+"\n");
 
+    // fetch the last int, i.e. last exception.
+    int lastExc = (res[ensz-4] << 24) | 
+         ((0xff & res[ensz-3]) << 16) | 
+         ((0xff & res[ensz-2]) << 8 ) | 
+          (0xff & res[ensz-1]);
+
+    // trailing forced exceptions are suppressed, 
+    // so the last exception should be what we assigned. 
+    assert lastExc==excValue;  
     assert cmp(data,sz,copy,sz)==true;
   }
 
@@ -87,18 +207,18 @@ public class TestPForUtil extends LuceneTestCase {
    */
   public void testAllDistribution() throws Exception {
     initRandom();
+    int sz = ForPostingsFormat.DEFAULT_BLOCK_SIZE;
+    int[] data = new int[sz];
     for (int i=0; i<=32; ++i) { // try to test every kinds of distribution
       double alpha=gen.nextDouble(); // rate of normal value
       for (int j=0; j<=32; ++j) {
-        tryDistribution(ForPostingsFormat.DEFAULT_BLOCK_SIZE,alpha,MASK[i],MASK[j]);
+        createDistribution(data,sz,alpha,MASK[i],MASK[j]);
+        tryCompressAndDecompress(data, sz);
       }
     }
   }
-  public void tryDistribution(int sz, double alpha, int masknorm, int maskexc) throws Exception {
+  public void createDistribution(int[] data, int sz, double alpha, int masknorm, int maskexc) {
     Integer[] buff= new Integer[sz];
-    int[] data = new int[sz];
-    byte[] res = new byte[4+sz*8];      // loosely upperbound
-    IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
     int i=0;
     for (; i<sz*alpha; ++i)
       buff[i]=gen.nextInt() & masknorm;
@@ -107,6 +227,10 @@ public class TestPForUtil extends LuceneTestCase {
     Collections.shuffle(Arrays.asList(buff),gen);
     for (i=0; i<sz; ++i)
       data[i] = buff[i];
+  }
+  public void tryCompressAndDecompress(final int[] data, int sz) throws Exception {
+    byte[] res = new byte[4+sz*8];      // loosely upperbound
+    IntBuffer resBuffer = ByteBuffer.wrap(res).asIntBuffer();
 
     int ensz = PForUtil.compress(data,sz,resBuffer);
     
@@ -160,6 +284,9 @@ public class TestPForUtil extends LuceneTestCase {
     }
     return hex.toString();
   }
+  static void eprintln(String format, Object... args) {
+    System.err.println(String.format(format,args)); 
+  }
   static void println(String format, Object... args) {
     System.out.println(String.format(format,args)); 
   }