KenLM 58da338b

2 files changed, 13 insertions, 7 deletions

diff --git a/klm/util/stream/io.cc b/klm/util/stream/io.cc
index c7ad2980..0459f706 100644
--- a/klm/util/stream/io.cc
+++ b/klm/util/stream/io.cc
@@ -29,15 +29,17 @@ void Read::Run(const ChainPosition &position) {
 void PRead::Run(const ChainPosition &position) {
   scoped_fd owner;
   if (own_) owner.reset(file_);
-  uint64_t size = SizeOrThrow(file_);
+  const uint64_t size = SizeOrThrow(file_);
   UTIL_THROW_IF(size % static_cast<uint64_t>(position.GetChain().EntrySize()), ReadSizeException, "File size " << file_ << " size is " << size << " not a multiple of " << position.GetChain().EntrySize());
-  std::size_t block_size = position.GetChain().BlockSize();
+  const std::size_t block_size = position.GetChain().BlockSize();
+  const uint64_t block_size64 = static_cast<uint64_t>(block_size);
   Link link(position);
   uint64_t offset = 0;
-  for (; offset + block_size < size; offset += block_size, ++link) {
+  for (; offset + block_size64 < size; offset += block_size64, ++link) {
     PReadOrThrow(file_, link->Get(), block_size, offset);
     link->SetValidSize(block_size);
   }
+  // size - offset is <= block_size, so it casts to 32-bit fine.
   if (size - offset) {
     PReadOrThrow(file_, link->Get(), size - offset, offset);
     link->SetValidSize(size - offset);
diff --git a/klm/util/stream/sort.hh b/klm/util/stream/sort.hh
index a86f160f..16aa6a03 100644
--- a/klm/util/stream/sort.hh
+++ b/klm/util/stream/sort.hh
@@ -365,10 +365,14 @@ template <class Compare> class BlockSorter {
         // Record the size of each block in a separate file.    
         offsets_->Append(link->ValidSize());
         void *end = static_cast<uint8_t*>(link->Get()) + link->ValidSize();
-        std::sort(
-            SizedIt(link->Get(), entry_size),
-            SizedIt(end, entry_size),
-            compare_);
+#if defined(_WIN32) || defined(_WIN64)
+        std::stable_sort
+#else
+        std::sort
+#endif
+          (SizedIt(link->Get(), entry_size),
+           SizedIt(end, entry_size),
+           compare_);
       }
       offsets_->FinishedAppending();
     }