From d34b107b91c2b91ef2fc7ee070f32a85b36c222e Mon Sep 17 00:00:00 2001
From: Ulrich Germann <ugermann@inf.ed.ac.uk>
Date: Mon, 8 Jun 2015 14:00:31 +0100
Subject: [PATCH] Initial check-in.

---
 .../TranslationModel/UG/mm/ug_prep_phrases.h  | 83 +++++++++++++++++++
 1 file changed, 83 insertions(+)
 create mode 100644 moses/TranslationModel/UG/mm/ug_prep_phrases.h

diff --git a/moses/TranslationModel/UG/mm/ug_prep_phrases.h b/moses/TranslationModel/UG/mm/ug_prep_phrases.h
new file mode 100644
index 000000000..421d6f090
--- /dev/null
+++ b/moses/TranslationModel/UG/mm/ug_prep_phrases.h
@@ -0,0 +1,83 @@
+// -*- mode: c++; tab-width: 2; indent-tabs-mode: nil; -*-
+// Functions for multi-threaded pre-fetching of phrase table entries
+// Author: Ulrich Germann
+
+#include "moses/TranslationModel/UG/generic/threading/ug_thread_pool.h"
+#include "moses/thread_safe_container.h"
+#include "ug_bitext.h"
+#include "ug_lru_cache.h"
+
+namespace Moses {
+namespace bitext { 
+
+template<typename Token> // , typename BITEXT>
+struct StatsCollector
+{
+  typedef    lru_cache::LRU_Cache< uint64_t, pstats  > hcache_t;
+  typedef ThreadSafeContainer<uint64_t, sptr<pstats> > pcache_t;
+  typedef                 map<uint64_t, sptr<pstats> > lcache_t;
+  iptr<Bitext<Token> const> bitext; // underlying bitext
+  sampling_method           method; // sampling method 
+  size_t               sample_size; // sample size 
+  sptr<SamplingBias const>    bias; // sampling bias
+  hcache_t*                 hcache; // "history" cache
+  pcache_t*                 pcache; // permanent cache
+  size_t                 pcache_th; // threshold for adding items to pcache 
+  sptr<lcache_t>            lcache; // local cache
+  ug::ThreadPool*            tpool; // thread pool to run jobs on 
+  
+  StatsCollector(iptr<Bitext<Token> > xbitext, 
+		 sptr<SamplingBias> const xbias) 
+    : method(ranked_sampling)
+    , sample_size(100)
+    , bias(xbias)
+    , hcache(NULL)
+    , pcache(NULL)
+    , pcache_th(10000)
+    , tpool(NULL)
+  { 
+    bitext = xbitext;
+  }
+
+  void
+  process(typename TSA<Token>::tree_iterator& m, 
+	  typename TSA<Token>::tree_iterator& r) 
+  {
+    if (!lcache) lcache.reset(new lcache_t);
+    if (m.down())
+      {
+	do 
+	  {
+	    if (!r.extend(m.getToken(-1)->id())) continue;
+	    this->process(m, r);
+	    uint64_t pid = r.getPid();
+	    sptr<pstats> stats;
+	    if (hcache) stats = hcache->get(pid); 
+	    if (!stats && pcache)
+	      {
+		sptr<pstats> const* foo = pcache->get(pid);
+		if (foo) stats = *foo; 
+	      }
+	    if (!stats) // need to sample
+	      {
+		BitextSampler<Token> s(bitext.get(), r, bias, sample_size, method);
+		stats = s.stats();
+		if (hcache) hcache->set(pid,stats);
+		if (pcache && r.ca() >= pcache_th) pcache->set(pid,stats);
+		if (tpool) tpool->add(s);
+		else s();
+	      }
+	    (*lcache)[pid] = stats;
+	    r.up();
+	  }
+	while (m.over());
+	m.up();
+      }
+  }
+};
+} // end of namespace bitext
+} // end of namespace Moses
+
+#if 0
+#endif
+	    // r.up();