Java UnsafeUtils.get4方法代码示例

import water.util.UnsafeUtils; //导入方法依赖的package包/类
/**
 * Optimized trim() method to operate across the entire CStrChunk buffer in one pass.
 * This mimics Java String.trim() by only considering characters of value
 * <code>'&#92;u0020'</code> or less as whitespace to be trimmed. This means that like
 * Java's String.trim() it ignores 16 of the 17 characters regarded as a space in UTF.
 *
 * NewChunk is the same size as the original, despite trimming.
 *
 * @param nc NewChunk to be filled with trimmed version of strings in this chunk
 * @return Filled NewChunk
 */
public NewChunk asciiTrim(NewChunk nc) {
  // copy existing data
  nc = this.inflate_impl(nc);
  //update offsets and byte array
  for(int i=0; i < _len; i++) {
    int j = 0;
    int off = UnsafeUtils.get4(_mem,(i<<2)+_OFF);
    if (off != NA) {
      //UTF chars will appear as negative values. In Java spec, space is any char 0x20 and lower
      while( _mem[_valstart+off+j] > 0 && _mem[_valstart+off+j] < 0x21) j++;
      if (j > 0) nc._is[i] = off + j;
      while( _mem[_valstart+off+j] != 0 ) j++; //Find end
      j--;
      while( _mem[_valstart+off+j] > 0 && _mem[_valstart+off+j] < 0x21) { //March back to find first non-space
        nc._ss[off+j] = 0; //Set new end
        j--;
      }
    }
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
/**
 * Optimized length() method for a buffer of only ASCII characters.
 * This is a straight byte count for each word in the chunk. The presence
 * of UTF-8 multi-byte characters would give incorrect results.
 *
 * @param nc NewChunk to be filled with lengths of strings in this chunk
 * @return Filled NewChunk
 */
public NewChunk asciiLength(NewChunk nc) {
  //pre-allocate since size is known
  nc._ls = MemoryManager.malloc8(_len);
  nc._xs = MemoryManager.malloc4(_len); // sadly, a waste
  // fill in lengths
  for(int i=0; i < _len; i++) {
    int off = UnsafeUtils.get4(_mem,(i<<2)+_OFF);
    int len = 0;
    if (off != NA) {
      while (_mem[_valstart + off + len] != 0) len++;
      nc.addNum(len, 0);
    } else nc.addNA();
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
/**
 * Optimized trim() method to operate across the entire CStrChunk buffer in one pass.
 * This mimics Java String.trim() by only considering characters of value
 * <code>'&#92;u0020'</code> or less as whitespace to be trimmed. This means that like
 * Java's String.trim() it ignores 16 of the 17 characters regarded as a space in UTF.
 *
 * NewChunk is the same size as the original, despite trimming.
 *
 * @param nc NewChunk to be filled with trimmed version of strings in this chunk
 * @return Filled NewChunk
 */
public NewChunk asciiTrim(NewChunk nc) {
  // copy existing data
  nc = this.extractRows(nc, 0,_len);
  //update offsets and byte array
  for(int i=0; i < _len; i++) {
    int j = 0;
    int off = UnsafeUtils.get4(_mem,idx(i));
    if (off != NA) {
      //UTF chars will appear as negative values. In Java spec, space is any char 0x20 and lower
      while( _mem[_valstart+off+j] > 0 && _mem[_valstart+off+j] < 0x21) j++;
      if (j > 0) nc.set_is(i,off + j);
      while( _mem[_valstart+off+j] != 0 ) j++; //Find end
      j--;
      while( _mem[_valstart+off+j] > 0 && _mem[_valstart+off+j] < 0x21) { //March back to find first non-space
        nc._ss[off+j] = 0; //Set new end
        j--;
      }
    }
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
/**
 * Optimized substring() method for a buffer of only ASCII characters.
 * The presence of UTF-8 multi-byte characters would give incorrect results
 * for the string length, which is required here.
 *
 * @param nc NewChunk to be filled with substrings in this chunk
 * @param startIndex The beginning index of the substring, inclusive
 * @param endIndex The ending index of the substring, exclusive
 * @return Filled NewChunk
 */
public NewChunk asciiSubstring(NewChunk nc, int startIndex, int endIndex) {
  // copy existing data
  nc = this.extractRows(nc, 0,_len);
  //update offsets and byte array
  for (int i = 0; i < _len; i++) {
    int off = UnsafeUtils.get4(_mem, idx(i));
    if (off != NA) {
      int len = 0;
      while (_mem[_valstart + off + len] != 0) len++; //Find length
      nc.set_is(i,startIndex < len ? off + startIndex : off + len);
      for (; len > endIndex - 1; len--) {
        nc._ss[off + len] = 0; //Set new end
      }
    }
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
/**
 * Optimized length() method for a buffer of only ASCII characters.
 * This is a straight byte count for each word in the chunk. The presence
 * of UTF-8 multi-byte characters would give incorrect results.
 *
 * @param nc NewChunk to be filled with lengths of strings in this chunk
 * @return Filled NewChunk
 */
public NewChunk asciiLength(NewChunk nc) {
  //pre-allocate since size is known
  nc.alloc_mantissa(_len);
  nc.alloc_exponent(_len); // sadly, a waste
  // fill in lengths
  for(int i=0; i < _len; i++) {
    int off = UnsafeUtils.get4(_mem,idx(i));
    int len = 0;
    if (off != NA) {
      while (_mem[_valstart + off + len] != 0) len++;
      nc.addNum(len, 0);
    } else nc.addNA();
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public NewChunk inflate_impl(NewChunk nc) {
  nc.set_sparseLen(0);
  nc.set_len(0);
  final int len = _len;
  for( int i=0; i<len; i++ ) {
    int res = UnsafeUtils.get4(_mem,(i<<2));
    if( res == _NA ) nc.addNA();
    else             nc.addNum(res,0);
  }
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public BufferedString atStr_impl(BufferedString bStr, int idx) {
  int off = UnsafeUtils.get4(_mem,(idx<<2)+_OFF);
  if( off == NA ) return null;
  int len = 0;
  while( _mem[_valstart+off+len] != 0 ) len++;
  return bStr.set(_mem,_valstart+off,len);
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public CStrChunk read_impl(AutoBuffer bb) {
  _mem = bb.bufClose();
  _start = -1;  _cidx = -1;
  _valstart = UnsafeUtils.get4(_mem,0);
  byte b = UnsafeUtils.get1(_mem,4);
  if (b == 0) _isAllASCII = false; else _isAllASCII = true;
  set_len((_valstart-_OFF)>>2);
  return this;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public NewChunk inflate_impl(NewChunk nc) {
  nc.set_len(_len);
  nc.set_sparseLen(sparseLen());
  nc._isAllASCII = _isAllASCII;
  nc._is = MemoryManager.malloc4(_len);
  for( int i = 0; i < _len; i++ )
    nc._is[i] = UnsafeUtils.get4(_mem,(i<<2)+_OFF);
  nc._sslen = _mem.length - _valstart;
  nc._ss = MemoryManager.malloc1(nc._sslen);
  System.arraycopy(_mem,_valstart,nc._ss,0,nc._sslen);
  return nc;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public CUDChunk read_impl(AutoBuffer bb) {
  _mem = bb.bufClose();
  _start = -1;  _cidx = -1;
  _len = UnsafeUtils.get4(_mem, 0);
  numUniques = UnsafeUtils.get4(_mem, 4);
  set_len(_len);
  return this;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
protected final long getIValue(int off){
  switch(_valsz){
    case 1: return _mem[off+ _ridsz]&0xFF;
    case 2: return UnsafeUtils.get2(_mem, off + _ridsz);
    case 4: return UnsafeUtils.get4(_mem, off + _ridsz);
    case 8: return UnsafeUtils.get8(_mem, off + _ridsz);
    default: throw H2O.fail();
 } 
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
static Compression guessCompressionMethod(byte [] bits) {
  // Look for ZIP magic
  if( bits.length > ZipFile.LOCHDR && UnsafeUtils.get4(bits, 0) == ZipFile.LOCSIG )
    return Compression.ZIP;
  if( bits.length > 2 && (UnsafeUtils.get2(bits,0)&0xffff) == GZIPInputStream.GZIP_MAGIC )
    return Compression.GZIP;
  return Compression.NONE;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override
protected final void initFromBytes() {
  _start = -1;  _cidx = -1;
  _len = UnsafeUtils.get4(_mem,0);
  int id_sz = _mem[4]&0xFF;
  _val_sz = _mem[5]&0xFF;
  _elem_sz = _val_sz + id_sz;
  _isNA = (0xFF&_mem[6]) == 1;
  _previousOffset = _OFF;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public double [] getDoubles(double [] vals, int from, int to, double NA){
  for(int i = from; i < to; i++) {
    int x = UnsafeUtils.get4(_mem, 4*i);
    vals[i-from] = (x == _NA)?NA:x;
  }
  return vals;
}
 

import water.util.UnsafeUtils; //导入方法依赖的package包/类
@Override public final void initFromBytes () {
  _start = -1;  _cidx = -1;
  set_len((_mem.length-_OFF) >> getSzLog());
  _bias = UnsafeUtils.get8 (_mem,0);
  int x = UnsafeUtils.get4(_mem,8);
  int szLog = UnsafeUtils.get4(_mem,12);
  _isDecimal = szLog < 0;
  _scale = PrettyPrint.pow10(1,_isDecimal?-x:x);
}