view release on metacpan or  search on metacpan

xgboost/dmlc-core/src/data/disk_row_iter.h  view on Meta::CPAN

/*!
 *  Copyright (c) 2015 by Contributors
 * \file basic_row_iter.h
 * \brief row based iterator that
 *   caches things into disk and then load segments
 * \author Tianqi Chen
 */
#ifndef DMLC_DATA_DISK_ROW_ITER_H_
#define DMLC_DATA_DISK_ROW_ITER_H_

#include <dmlc/io.h>
#include <dmlc/logging.h>
#include <dmlc/data.h>
#include <dmlc/timer.h>
#include <dmlc/threadediter.h>
#include <algorithm>
#include <string>
#include "./row_block.h"
#include "./libsvm_parser.h"

#if DMLC_ENABLE_STD_THREAD
namespace dmlc {
namespace data {
/*!
 * \brief basic set of row iterators that provides
 * \tparam IndexType the type of index we are using
 */
template<typename IndexType>
class DiskRowIter: public RowBlockIter<IndexType> {
 public:
  // page size 64MB
  static const size_t kPageSize = 64UL << 20UL;
  /*!
   * \brief disk row iterator constructor
   * \param parser parser used to generate this

   */
  explicit DiskRowIter(Parser<IndexType> *parser,
                       const char *cache_file,
                       bool reuse_cache)
      : cache_file_(cache_file), fi_(NULL) {
    if (reuse_cache) {
      if (!TryLoadCache()) {
        this->BuildCache(parser);
        CHECK(TryLoadCache())
            << "failed to build cache file " << cache_file;
      }
    } else {
      this->BuildCache(parser);
      CHECK(TryLoadCache())
          << "failed to build cache file " << cache_file;
    }
    delete parser;
  }
  virtual ~DiskRowIter(void) {
    iter_.Destroy();
    delete fi_;
  }
  virtual void BeforeFirst(void) {
    iter_.BeforeFirst();
  }
  virtual bool Next(void) {
    if (iter_.Next()) {
      row_ = iter_.Value().GetBlock();
      return true;
    } else {
      return false;
    }
  }
  virtual const RowBlock<IndexType> &Value(void) const {
    return row_;
  }
  virtual size_t NumCol(void) const {
    return num_col_;