Allow compilation with Visual Studio; add flag for output file instead of stdout; document flags in usage arg; allow forced align to produce same output format as train+align.

CMakeLists.txt

@@ -7,9 +7,10 @@ include_directories(${CMAKE_CURRENT_SOURCE_DIR})
 find_package(SparseHash)
 if(SPARSEHASH_FOUND)
   add_definitions(-DHAVE_SPARSEHASH)
+  include_directories(${SPARSEHASH_INCLUDE_DIR})
 endif(SPARSEHASH_FOUND)
 
-find_package(OpenMP QUIET)
+find_package(OpenMP)
 if (OPENMP_FOUND)
   set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
 endif(OPENMP_FOUND)

README.md

@@ -49,6 +49,22 @@ These can be symmetrized using the included `atools` command using a variety of
 
     ./atools -i forward.align -j reverse.align -c grow-diag-final-and
 
+
+## Compiling for windows
+
+To build 64-bit binaries with libsparsehash on Windows using Visual Studio 2017, the following recipe works:
+
+
+    cd d:\src
+    git clone https://github.com/clab/fast_align.git
+    git clone https://github.com/sparsehash/sparsehash.git
+    cd d:\src\fast_align
+    mkdir build64
+    cd build64
+    cmake -G "Visual Studio 15 2017 Win64" -D SPARSEHASH_INCLUDE_DIR=d:\src\sparsehash\src ..
+    cmake --build build64 --config Release
+
+
 ## Output
 
 `fast_align` produces outputs in the widely-used `i-j` “Pharaoh format,” where a pair `i-j` indicates that the <i>i</i>th word (zero-indexed) of the left language (by convention, the *source* language) is aligned to the <i>j</i>th word of the right sentence (by convention, the *target* language). For example, a good alignment of the above German–English corpus would be:

src/atools.cc

@@ -6,7 +6,11 @@
 #include <queue>
 #include <map>
 #include <set>
+#ifndef _MSC_VER
 #include <getopt.h>
+#else
+#include "getopt.h"
+#endif
 
 #include "alignment_io.h"
 
@@ -16,22 +20,27 @@ struct option options[] = {
     {"input_1",                 required_argument, 0,                  'i'},
     {"input_2",                 required_argument, 0,                  'j'},
     {"command",                 required_argument, 0,                  'c'},
+    {"outfile",                 required_argument, 0,                  'o'},
     {0,0,0,0}
 };
 
 string input_1;
 string input_2;
 string command;
+ofstream outfileStream;
+ostream* outstream = &cout;
+
 
 bool InitCommandLine(int argc, char** argv) {
   while (1) {
     int oi;
-    int c = getopt_long(argc, argv, "i:j:c:", options, &oi);
+    int c = getopt_long(argc, argv, "i:j:c:o:", options, &oi);
     if (c == -1) break;
     switch(c) {
       case 'i': input_1 = optarg; break;
       case 'j': input_2 = optarg; break;
       case 'c': command = optarg; break;
+      case 'o': outfileStream.open(optarg); outstream = &outfileStream; break;
       default: return false;
     }
   }
@@ -308,7 +317,7 @@ int main(int argc, char **argv) {
   AddCommand<GDFACommand>();
   AddCommand<FMeasureCommand>();
   if (!InitCommandLine(argc, argv)) {
-    cerr << "Usage: " << argv[0] << " -c COMMAND -i FILE1.AL [-j FILE2.AL]\n";
+    cerr << "Usage: " << argv[0] << " -c COMMAND -i FILE1.AL [-j FILE2.AL] [-o OUTPATH]\n";
     cerr << "Valid options for COMMAND:";
     for (auto it : commands)
       cerr << ' ' << it.first;
@@ -368,7 +377,7 @@ int main(int argc, char **argv) {
     }
 
     if (cmd.Result() == 1) {
-      AlignmentIO::SerializePharaohFormat(*out, &cout);
+      AlignmentIO::SerializePharaohFormat(*out, outstream);
     }
   }
   if (cmd.Result() == 2)

src/corpus.h

@@ -22,15 +22,15 @@ class Dict {
     return (x == ' ' || x == '\t');
   }
 
-  inline void ConvertWhitespaceDelimitedLine(const std::string& line, std::vector<unsigned>* out) {
+  inline void ConvertWhitespaceDelimitedLine(const std::string& line, std::vector<unsigned>* out, bool frozen=false) {
     size_t cur = 0;
     size_t last = 0;
     int state = 0;
     out->clear();
     while(cur < line.size()) {
       if (is_ws(line[cur++])) {
         if (state == 0) continue;
-        out->push_back(Convert(line.substr(last, cur - last - 1)));
+        out->push_back(Convert(line.substr(last, cur - last - 1), frozen));
         state = 0;
       } else {
         if (state == 1) continue;