From deb555e5ab40a62738269050b43b412335d4b66a Mon Sep 17 00:00:00 2001
From: Patrick Simianer <p@simianer.de>
Date: Tue, 12 Apr 2016 10:07:48 +0200
Subject: extractor: gzip compressed grammars

---
 extractor/extract.cc       |  2 +-
 extractor/run_extractor.cc | 13 ++++++++-----
 2 files changed, 9 insertions(+), 6 deletions(-)

(limited to 'extractor')

diff --git a/extractor/extract.cc b/extractor/extract.cc
index 08f209cc..b16a4e1c 100644
--- a/extractor/extract.cc
+++ b/extractor/extract.cc
@@ -14,7 +14,6 @@
   const unsigned omp_get_num_threads() { return 1; }
 #endif
 
-#include "filelib.h"
 #include "alignment.h"
 #include "data_array.h"
 #include "features/count_source_target.h"
@@ -25,6 +24,7 @@
 #include "features/max_lex_target_given_source.h"
 #include "features/sample_source_count.h"
 #include "features/target_given_source_coherent.h"
+#include "filelib.h"
 #include "grammar.h"
 #include "grammar_extractor.h"
 #include "precomputation.h"
diff --git a/extractor/run_extractor.cc b/extractor/run_extractor.cc
index 00564a36..81d0d8be 100644
--- a/extractor/run_extractor.cc
+++ b/extractor/run_extractor.cc
@@ -24,6 +24,7 @@
 #include "features/max_lex_target_given_source.h"
 #include "features/sample_source_count.h"
 #include "features/target_given_source_coherent.h"
+#include "filelib.h"
 #include "grammar.h"
 #include "grammar_extractor.h"
 #include "precomputation.h"
@@ -41,8 +42,8 @@ using namespace extractor;
 using namespace features;
 
 // Returns the file path in which a given grammar should be written.
-fs::path GetGrammarFilePath(const fs::path& grammar_path, int file_number) {
-  string file_name = "grammar." + to_string(file_number);
+fs::path GetGrammarFilePath(const fs::path& grammar_path, int file_number, bool use_zip) {
+  string file_name = "grammar." + to_string(file_number) + (use_zip ? ".gz" : "");
   return grammar_path / file_name;
 }
 
@@ -61,6 +62,7 @@ int main(int argc, char** argv) {
     ("bitext,b", po::value<string>(), "Parallel text (source ||| target)")
     ("alignment,a", po::value<string>()->required(), "Bitext word alignment")
     ("grammars,g", po::value<string>()->required(), "Grammars output path")
+    ("gzip,z", "Gzip grammars")
     ("threads,t", po::value<int>()->default_value(1), threads_option.c_str())
     ("frequent", po::value<int>()->default_value(100),
         "Number of precomputed frequent patterns")
@@ -205,6 +207,7 @@ int main(int argc, char** argv) {
       vm["max_rule_symbols"].as<int>(),
       vm["max_samples"].as<int>(),
       vm["tight_phrases"].as<bool>());
+  const bool use_zip = vm.count("gzip");
 
   // Creates the grammars directory if it doesn't exist.
   fs::path grammar_path = vm["grammars"].as<string>();
@@ -239,12 +242,12 @@ int main(int argc, char** argv) {
     }
     Grammar grammar = extractor.GetGrammar(
         sentences[i], blacklisted_sentence_ids);
-    ofstream output(GetGrammarFilePath(grammar_path, i).c_str());
-    output << grammar;
+    WriteFile wf(GetGrammarFilePath(grammar_path, i, use_zip).c_str());
+    *wf.stream() << grammar;
   }
 
   for (size_t i = 0; i < sentences.size(); ++i) {
-    cout << "<seg grammar=" << GetGrammarFilePath(grammar_path, i) << " id=\""
+    cout << "<seg grammar=" << GetGrammarFilePath(grammar_path, i, use_zip) << " id=\""
          << i << "\"> " << sentences[i] << " </seg> " << suffixes[i] << endl;
   }
 
-- 
cgit v1.2.3