KenLM a6d57501dcac95a31719a8628f6cbd288f6741e2 including Marcin's fixed pruning

6 files changed, 29 insertions, 10 deletions

diff --git a/util/exception.cc b/util/exception.cc
index 083bac20d..eaf4cbc21 100644
--- a/util/exception.cc
+++ b/util/exception.cc
@@ -51,6 +51,11 @@ void Exception::SetLocation(const char *file, unsigned int line, const char *fun
 }
 
 namespace {
+
+#ifdef __GNUC__
+const char *HandleStrerror(int ret, const char *buf) __attribute__ ((unused));
+const char *HandleStrerror(const char *ret, const char * /*buf*/) __attribute__ ((unused));
+#endif
 // At least one of these functions will not be called.
 #ifdef __clang__
 #pragma clang diagnostic push
diff --git a/util/fake_ofstream.hh b/util/fake_ofstream.hh
index eefb1edce..987fa8015 100644
--- a/util/fake_ofstream.hh
+++ b/util/fake_ofstream.hh
@@ -17,16 +17,15 @@
 namespace util {
 class FakeOFStream {
   public:
-    static const std::size_t kOutBuf = 1048576;
-
     // Does not take ownership of out.
     // Allows default constructor, but must call SetFD.
-    explicit FakeOFStream(int out = -1)
-      : buf_(util::MallocOrThrow(kOutBuf)),
-        builder_(static_cast<char*>(buf_.get()), kOutBuf),
+    explicit FakeOFStream(int out = -1, std::size_t buffer_size = 1048576)
+      : buf_(util::MallocOrThrow(buffer_size)),
+        builder_(static_cast<char*>(buf_.get()), buffer_size),
         // Mostly the default but with inf instead.  And no flags.
         convert_(double_conversion::DoubleToStringConverter::NO_FLAGS, "inf", "NaN", 'e', -6, 21, 6, 0),
-        fd_(out) {}
+        fd_(out),
+        buffer_size_(buffer_size) {}
 
     ~FakeOFStream() {
       if (buf_.get()) Flush();
@@ -51,7 +50,7 @@ class FakeOFStream {
     }
 
     FakeOFStream &operator<<(StringPiece str) {
-      if (str.size() > kOutBuf) {
+      if (str.size() > buffer_size_) {
         Flush();
         util::WriteOrThrow(fd_, str.data(), str.size());
       } else {
@@ -98,6 +97,7 @@ class FakeOFStream {
     double_conversion::StringBuilder builder_;
     double_conversion::DoubleToStringConverter convert_;
     int fd_;
+    const std::size_t buffer_size_;
 };
 
 } // namespace
diff --git a/util/stream/chain.cc b/util/stream/chain.cc
index 4596af7ae..ce29e428c 100644
--- a/util/stream/chain.cc
+++ b/util/stream/chain.cc
@@ -128,7 +128,7 @@ Link::~Link() {
   if (current_) {
     // Probably an exception unwinding.  
     std::cerr << "Last input should have been poison." << std::endl;
-//    abort();
+    // abort();
   } else {
     if (!poisoned_) {
       // Poison is a block whose memory pointer is NULL.
diff --git a/util/stream/io.cc b/util/stream/io.cc
index c64004c0b..fa8467abd 100644
--- a/util/stream/io.cc
+++ b/util/stream/io.cc
@@ -70,6 +70,8 @@ void PWriteAndRecycle::Run(const ChainPosition &position) {
     offset += link->ValidSize();
     link->SetValidSize(block_size);
   }
+  // Trim file to size.
+  util::ResizeOrThrow(file_, offset);
 }
 
 } // namespace stream
diff --git a/util/usage.cc b/util/usage.cc
index 2a4aa47d4..a597300e3 100644
--- a/util/usage.cc
+++ b/util/usage.cc
@@ -68,6 +68,18 @@ Wall GetWall() {
 }
 #endif
 
+// gcc possible-unused function flags
+#ifdef __GNUC__
+double Subtract(time_t first, time_t second) __attribute__ ((unused));
+double DoubleSec(time_t tv) __attribute__ ((unused));
+#if !defined(_WIN32) && !defined(_WIN64)
+double Subtract(const struct timeval &first, const struct timeval &second) __attribute__ ((unused));
+double Subtract(const struct timespec &first, const struct timespec &second) __attribute__ ((unused));
+double DoubleSec(const struct timeval &tv) __attribute__ ((unused));
+double DoubleSec(const struct timespec &tv) __attribute__ ((unused));
+#endif
+#endif
+
 // Some of these functions are only used on some platforms.
 #ifdef __clang__
 #pragma clang diagnostic push
diff --git a/util/usage.hh b/util/usage.hh
index 5407f5a54..e578b0a65 100644
--- a/util/usage.hh
+++ b/util/usage.hh
@@ -13,9 +13,9 @@ double WallTime();
 void PrintUsage(std::ostream &to);
 
 // Determine how much physical memory there is.  Return 0 on failure.
-::uint64_t GuessPhysicalMemory();
+uint64_t GuessPhysicalMemory();
 
 // Parse a size like unix sort.  Sadly, this means the default multiplier is K.
-::uint64_t ParseSize(const std::string &arg);
+uint64_t ParseSize(const std::string &arg);
 } // namespace util
 #endif // UTIL_USAGE_H