KenLM update. Remove a couple of segfaults for weird input. Other oddball stuff.

6 files changed, 25 insertions, 18 deletions

diff --git a/klm/util/file.cc b/klm/util/file.cc
index 6a3885a7..98f13983 100644
--- a/klm/util/file.cc
+++ b/klm/util/file.cc
@@ -44,6 +44,16 @@ int OpenReadOrThrow(const char *name) {
   return ret;
 }
 
+int CreateOrThrow(const char *name) {
+  int ret;
+#if defined(_WIN32) || defined(_WIN64)
+  UTIL_THROW_IF(-1 == (ret = _open(name, _O_CREAT | _O_TRUNC | _O_RDWR, _S_IREAD | _S_IWRITE)), ErrnoException, "while creating " << name);
+#else
+  UTIL_THROW_IF(-1 == (ret = open(name, O_CREAT | O_TRUNC | O_RDWR, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH)), ErrnoException, "while creating " << name);
+#endif
+  return ret;
+}
+
 uint64_t SizeFile(int fd) {
 #if defined(_WIN32) || defined(_WIN64)
   __int64 ret = _filelengthi64(fd);
diff --git a/klm/util/file.hh b/klm/util/file.hh
index 5c57e2a9..8af1ff4f 100644
--- a/klm/util/file.hh
+++ b/klm/util/file.hh
@@ -65,7 +65,10 @@ class scoped_FILE {
     std::FILE *file_;
 };
 
+// Open for read only.  
 int OpenReadOrThrow(const char *name);
+// Create file if it doesn't exist, truncate if it does.  Opened for write.   
+int CreateOrThrow(const char *name);
 
 // Return value for SizeFile when it can't size properly.  
 const uint64_t kBadSize = (uint64_t)-1;
@@ -91,10 +94,13 @@ class TempMaker {
   public:
     explicit TempMaker(const std::string &prefix);
 
+    // These will already be unlinked for you.  
     int Make() const;
-
     std::FILE *MakeFile() const;
 
+    // This will force you to close the fd instead of leaving it open.  
+    std::string Name(scoped_fd &opened) const;
+
   private:
     std::string base_;
 };
diff --git a/klm/util/file_piece.cc b/klm/util/file_piece.cc
index a205995a..19a68728 100644
--- a/klm/util/file_piece.cc
+++ b/klm/util/file_piece.cc
@@ -27,7 +27,7 @@ ParseNumberException::ParseNumberException(StringPiece value) throw() {
 #ifdef HAVE_ZLIB
 GZException::GZException(gzFile file) {
   int num;
-  *this << gzerror( file, &num) << " from zlib";
+  *this << gzerror(file, &num) << " from zlib";
 }
 #endif // HAVE_ZLIB
 
diff --git a/klm/util/have.hh b/klm/util/have.hh
index b8181e99..1d76a7fc 100644
--- a/klm/util/have.hh
+++ b/klm/util/have.hh
@@ -13,7 +13,7 @@
 #endif
 
 #ifndef HAVE_BOOST
-#define HAVE_BOOST
+//#define HAVE_BOOST
 #endif
 
 #ifndef HAVE_THREADS
diff --git a/klm/util/mmap.cc b/klm/util/mmap.cc
index 576fd4cc..bc9e3f81 100644
--- a/klm/util/mmap.cc
+++ b/klm/util/mmap.cc
@@ -19,8 +19,8 @@
 #include <windows.h>
 #include <io.h>
 #else
-#include <unistd.h>
 #include <sys/mman.h>
+#include <unistd.h>
 #endif
 
 namespace util {
@@ -171,20 +171,6 @@ void *MapZeroedWrite(int fd, std::size_t size) {
   return MapOrThrow(size, true, kFileFlags, false, fd, 0);
 }
 
-namespace {
-
-int CreateOrThrow(const char *name) {
-  int ret;
-#if defined(_WIN32) || defined(_WIN64)
-  UTIL_THROW_IF(-1 == (ret = _open(name, _O_CREAT | _O_TRUNC | _O_RDWR, _S_IREAD | _S_IWRITE)), ErrnoException, "while creating " << name);
-#else
-  UTIL_THROW_IF(-1 == (ret = open(name, O_CREAT | O_TRUNC | O_RDWR, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH)), ErrnoException, "while creating " << name);
-#endif
-  return ret;
-}
-
-} // namespace
-
 void *MapZeroedWrite(const char *name, std::size_t size, scoped_fd &file) {
   file.reset(CreateOrThrow(name));
   try {
diff --git a/klm/util/string_piece.hh b/klm/util/string_piece.hh
index 5de053aa..be6a643d 100644
--- a/klm/util/string_piece.hh
+++ b/klm/util/string_piece.hh
@@ -85,6 +85,11 @@ U_NAMESPACE_BEGIN
 #include <string>
 #include <string.h>
 
+#ifdef WIN32
+#undef max
+#undef min
+#endif
+
 class StringPiece {
  public:
   typedef size_t size_type;