diff options
author | Avneesh Saluja <asaluja@gmail.com> | 2013-03-28 18:28:16 -0700 |
---|---|---|
committer | Avneesh Saluja <asaluja@gmail.com> | 2013-03-28 18:28:16 -0700 |
commit | 3d8d656fa7911524e0e6885647173474524e0784 (patch) | |
tree | 81b1ee2fcb67980376d03f0aa48e42e53abff222 /klm/lm/builder/joint_order.hh | |
parent | be7f57fdd484e063775d7abf083b9fa4c403b610 (diff) | |
parent | 96fedabebafe7a38a6d5928be8fff767e411d705 (diff) |
fixed conflicts
Diffstat (limited to 'klm/lm/builder/joint_order.hh')
-rw-r--r-- | klm/lm/builder/joint_order.hh | 43 |
1 files changed, 43 insertions, 0 deletions
diff --git a/klm/lm/builder/joint_order.hh b/klm/lm/builder/joint_order.hh new file mode 100644 index 00000000..b5620144 --- /dev/null +++ b/klm/lm/builder/joint_order.hh @@ -0,0 +1,43 @@ +#ifndef LM_BUILDER_JOINT_ORDER__ +#define LM_BUILDER_JOINT_ORDER__ + +#include "lm/builder/multi_stream.hh" +#include "lm/lm_exception.hh" + +#include <string.h> + +namespace lm { namespace builder { + +template <class Callback, class Compare> void JointOrder(const ChainPositions &positions, Callback &callback) { + // Allow matching to reference streams[-1]. + NGramStreams streams_with_dummy; + streams_with_dummy.InitWithDummy(positions); + NGramStream *streams = streams_with_dummy.begin() + 1; + + unsigned int order; + for (order = 0; order < positions.size() && streams[order]; ++order) {} + assert(order); // should always have <unk>. + unsigned int current = 0; + while (true) { + // Does the context match the lower one? + if (!memcmp(streams[static_cast<int>(current) - 1]->begin(), streams[current]->begin() + Compare::kMatchOffset, sizeof(WordIndex) * current)) { + callback.Enter(current, *streams[current]); + // Transition to looking for extensions. + if (++current < order) continue; + } + // No extension left. + while(true) { + assert(current > 0); + --current; + callback.Exit(current, *streams[current]); + if (++streams[current]) break; + UTIL_THROW_IF(order != current + 1, FormatLoadException, "Detected n-gram without matching suffix"); + order = current; + if (!order) return; + } + } +} + +}} // namespaces + +#endif // LM_BUILDER_JOINT_ORDER__ |