From 6b780718b1af069992f4f7311c1cb753c8a68d05 Mon Sep 17 00:00:00 2001 From: alzobnin <alzobnin@yandex-team.com> Date: Thu, 15 Dec 2022 18:44:25 +0300 Subject: Restrict max length of learned prefixes and fix solar codec --- library/cpp/codecs/greedy_dict/gd_builder.cpp | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'library/cpp/codecs/greedy_dict/gd_builder.cpp') diff --git a/library/cpp/codecs/greedy_dict/gd_builder.cpp b/library/cpp/codecs/greedy_dict/gd_builder.cpp index 844e07d5a0..33e104926e 100644 --- a/library/cpp/codecs/greedy_dict/gd_builder.cpp +++ b/library/cpp/codecs/greedy_dict/gd_builder.cpp @@ -53,7 +53,7 @@ namespace NGreedyDict { Current->SetModelP(); } - ui32 TDictBuilder::BuildNextGeneration(ui32 maxent) { + ui32 TDictBuilder::BuildNextGeneration(ui32 maxent, ui32 maxlen) { TAutoPtr<TEntrySet> newset = new TEntrySet; newset->InitWithAlpha(); maxent -= newset->size(); @@ -86,7 +86,7 @@ namespace NGreedyDict { const TEntry& next = set.Get(Next(it->first)); float modelp = ModelP(prev.Count, next.Count, total); ui32 cnt = it->second; - if (cnt > mincnt && StatTest(test, modelp, cnt, total) > minpval) + if (cnt > mincnt && StatTest(test, modelp, cnt, total) > minpval && prev.Len() + next.Len() <= maxlen) Candidates.push_back(TCandidate(-Score(score, prev.Len() + next.Len(), modelp, cnt, total), it->first)); } } @@ -113,7 +113,7 @@ namespace NGreedyDict { return deletions + additions; } - ui32 TDictBuilder::Build(ui32 maxentries, ui32 maxiters, ui32 mindiff) { + ui32 TDictBuilder::Build(ui32 maxentries, ui32 maxiters, ui32 maxlen, ui32 mindiff) { /* size_t totalsz = 0; for (auto it : Input) totalsz += it.size();*/ @@ -128,7 +128,7 @@ namespace NGreedyDict { Clog << Sprintf("%-110s RSS=%" PRIu32 "M", mess.data(), (ui32)(TRusage::Get().MaxRss >> 20)) << Endl; } - ui32 diff = BuildNextGeneration(maxentries); + ui32 diff = BuildNextGeneration(maxentries, maxlen); if (Current->size() == maxentries && diff < mindiff) break; -- cgit v1.2.3