From d884099e0db8b4510847ec106b59ef7dca3c245b Mon Sep 17 00:00:00 2001 From: Kenneth Heafield Date: Fri, 18 Jan 2013 17:12:51 +0000 Subject: KenLM dffafbf with lmplz source (but not built) --- klm/util/stream/stream.hh | 74 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 74 insertions(+) create mode 100644 klm/util/stream/stream.hh (limited to 'klm/util/stream/stream.hh') diff --git a/klm/util/stream/stream.hh b/klm/util/stream/stream.hh new file mode 100644 index 00000000..6ff45b82 --- /dev/null +++ b/klm/util/stream/stream.hh @@ -0,0 +1,74 @@ +#ifndef UTIL_STREAM_STREAM__ +#define UTIL_STREAM_STREAM__ + +#include "util/stream/chain.hh" + +#include + +#include +#include + +namespace util { +namespace stream { + +class Stream : boost::noncopyable { + public: + Stream() : current_(NULL), end_(NULL) {} + + void Init(const ChainPosition &position) { + entry_size_ = position.GetChain().EntrySize(); + block_size_ = position.GetChain().BlockSize(); + block_it_.Init(position); + StartBlock(); + } + + explicit Stream(const ChainPosition &position) { + Init(position); + } + + operator bool() const { return current_ != NULL; } + bool operator!() const { return current_ == NULL; } + + const void *Get() const { return current_; } + void *Get() { return current_; } + + void Poison() { + block_it_->SetValidSize(current_ - static_cast(block_it_->Get())); + ++block_it_; + block_it_.Poison(); + } + + Stream &operator++() { + assert(*this); + assert(current_ < end_); + current_ += entry_size_; + if (current_ == end_) { + ++block_it_; + StartBlock(); + } + return *this; + } + + private: + void StartBlock() { + for (; block_it_ && !block_it_->ValidSize(); ++block_it_) {} + current_ = static_cast(block_it_->Get()); + end_ = current_ + block_it_->ValidSize(); + } + + uint8_t *current_, *end_; + + std::size_t entry_size_; + std::size_t block_size_; + + Link block_it_; +}; + +inline Chain &operator>>(Chain &chain, Stream &stream) { + stream.Init(chain.Add()); + return chain; +} + +} // namespace stream +} // namespace util +#endif // UTIL_STREAM_STREAM__ -- cgit v1.2.3