Merge branch 'master' of github.com:pks/cdec-dtrain

11 files changed, 18 insertions, 15 deletions

diff --git a/klm/lm/Makefile.am b/klm/lm/Makefile.am
index f15cbd77..48b0ba34 100644
--- a/klm/lm/Makefile.am
+++ b/klm/lm/Makefile.am
@@ -1,8 +1,11 @@
-bin_PROGRAMS = build_binary
+bin_PROGRAMS = build_binary ngram_query
 
-build_binary_SOURCES = build_binary.cc
+build_binary_SOURCES = build_binary_main.cc
 build_binary_LDADD = libklm.a ../util/libklm_util.a ../util/double-conversion/libklm_util_double.a -lz
 
+ngram_query_SOURCES = query_main.cc
+ngram_query_LDADD = libklm.a ../util/libklm_util.a ../util/double-conversion/libklm_util_double.a -lz
+
 #noinst_PROGRAMS = \
 #  ngram_test
 #TESTS = ngram_test
@@ -46,7 +49,6 @@ libklm_a_SOURCES = \
   lm_exception.cc \
 	quantize.cc \
   model.cc \
-  ngram_query.cc \
   read_arpa.cc \
   search_hashed.cc \
   search_trie.cc \
diff --git a/klm/lm/build_binary.cc b/klm/lm/build_binary_main.cc
index ab2c0c32..ab2c0c32 100644
--- a/klm/lm/build_binary.cc
+++ b/klm/lm/build_binary_main.cc
diff --git a/klm/lm/builder/Makefile.am b/klm/lm/builder/Makefile.am
index b5c147fd..317e03ce 100644
--- a/klm/lm/builder/Makefile.am
+++ b/klm/lm/builder/Makefile.am
@@ -1,7 +1,7 @@
 bin_PROGRAMS = builder
 
 builder_SOURCES = \
-  main.cc \
+  lmplz_main.cc \
   adjust_counts.cc \
   adjust_counts.hh \
   corpus_count.cc \
diff --git a/klm/lm/builder/discount.hh b/klm/lm/builder/discount.hh
index 754fb20d..4d0aa4fd 100644
--- a/klm/lm/builder/discount.hh
+++ b/klm/lm/builder/discount.hh
@@ -3,7 +3,7 @@
 
 #include <algorithm>
 
-#include <inttypes.h>
+#include <stdint.h>
 
 namespace lm {
 namespace builder {
diff --git a/klm/lm/builder/main.cc b/klm/lm/builder/lmplz_main.cc
index 90b9dca2..90b9dca2 100644
--- a/klm/lm/builder/main.cc
+++ b/klm/lm/builder/lmplz_main.cc
diff --git a/klm/lm/filter/main.cc b/klm/lm/filter/filter_main.cc
index c42243e2..1a4ba84f 100644
--- a/klm/lm/filter/main.cc
+++ b/klm/lm/filter/filter_main.cc
@@ -53,7 +53,7 @@ void DisplayHelp(const char *name) {
     "    stream i.e. /dev/stdout\n";
 }
 
-typedef enum {MODE_COPY, MODE_SINGLE, MODE_MULTIPLE, MODE_UNION} FilterMode;
+typedef enum {MODE_COPY, MODE_SINGLE, MODE_MULTIPLE, MODE_UNION, MODE_UNSET} FilterMode;
 typedef enum {FORMAT_ARPA, FORMAT_COUNT} Format;
 
 struct Config {
@@ -162,19 +162,19 @@ int main(int argc, char *argv[]) {
     return 1;
   }
 
-  // I used to have boost::program_options, but some users didn't want to compile boost.  
+  // I used to have boost::program_options, but some users didn't want to compile boost.
   lm::Config config;
-  boost::optional<lm::FilterMode> mode;
+  config.mode = lm::MODE_UNSET;
   for (int i = 1; i < argc - 2; ++i) {
     const char *str = argv[i];
     if (!std::strcmp(str, "copy")) {
-      mode = lm::MODE_COPY;
+      config.mode = lm::MODE_COPY;
     } else if (!std::strcmp(str, "single")) {
-      mode = lm::MODE_SINGLE;
+      config.mode = lm::MODE_SINGLE;
     } else if (!std::strcmp(str, "multiple")) {
-      mode = lm::MODE_MULTIPLE;
+      config.mode = lm::MODE_MULTIPLE;
     } else if (!std::strcmp(str, "union")) {
-      mode = lm::MODE_UNION;
+      config.mode = lm::MODE_UNION;
     } else if (!std::strcmp(str, "phrase")) {
       config.phrase = true;
     } else if (!std::strcmp(str, "context")) {
@@ -203,13 +203,12 @@ int main(int argc, char *argv[]) {
     }
   }
   
-  if (!mode) {
+  if (config.mode == lm::MODE_UNSET) {
     lm::DisplayHelp(argv[0]);
     return 1;
   }
-  config.mode = *mode;
 
-  if (config.phrase && config.mode != lm::MODE_UNION && mode != lm::MODE_MULTIPLE) {
+  if (config.phrase && config.mode != lm::MODE_UNION && config.mode != lm::MODE_MULTIPLE) {
     std::cerr << "Phrase constraint currently only works in multiple or union mode.  If you really need it for single, put everything on one line and use union." << std::endl;
     return 1;
   }
diff --git a/klm/lm/filter/phrase.hh b/klm/lm/filter/phrase.hh
index 07479dea..b4edff41 100644
--- a/klm/lm/filter/phrase.hh
+++ b/klm/lm/filter/phrase.hh
@@ -57,6 +57,7 @@ class Substrings {
     LM_FILTER_PHRASE_METHOD(Right, right)
     LM_FILTER_PHRASE_METHOD(Phrase, phrase)
 
+#pragma GCC diagnostic ignored "-Wuninitialized" // end != finish so there's always an initialization
     // sentence_id must be non-decreasing.  Iterators are over words in the phrase.  
     template <class Iterator> void AddPhrase(unsigned int sentence_id, const Iterator &begin, const Iterator &end) {
       // Iterate over all substrings.  
diff --git a/klm/lm/filter/vocab.hh b/klm/lm/filter/vocab.hh
index e2b6adff..7f0fadaa 100644
--- a/klm/lm/filter/vocab.hh
+++ b/klm/lm/filter/vocab.hh
@@ -5,6 +5,7 @@
 
 #include "util/multi_intersection.hh"
 #include "util/string_piece.hh"
+#include "util/string_piece_hash.hh"
 #include "util/tokenize_piece.hh"
 
 #include <boost/noncopyable.hpp>
diff --git a/klm/lm/fragment.cc b/klm/lm/fragment_main.cc
index 0267cd4e..0267cd4e 100644
--- a/klm/lm/fragment.cc
+++ b/klm/lm/fragment_main.cc
diff --git a/klm/lm/max_order.cc b/klm/lm/kenlm_max_order_main.cc
index 94221201..94221201 100644
--- a/klm/lm/max_order.cc
+++ b/klm/lm/kenlm_max_order_main.cc
diff --git a/klm/lm/ngram_query.cc b/klm/lm/query_main.cc
index 49757d9a..49757d9a 100644
--- a/klm/lm/ngram_query.cc
+++ b/klm/lm/query_main.cc