From d884099e0db8b4510847ec106b59ef7dca3c245b Mon Sep 17 00:00:00 2001
From: Kenneth Heafield <github@kheafield.com>
Date: Fri, 18 Jan 2013 17:12:51 +0000
Subject: KenLM dffafbf with lmplz source (but not built)

---
 klm/util/stream/sort_test.cc | 62 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)
 create mode 100644 klm/util/stream/sort_test.cc

(limited to 'klm/util/stream/sort_test.cc')

diff --git a/klm/util/stream/sort_test.cc b/klm/util/stream/sort_test.cc
new file mode 100644
index 00000000..fd7705cd
--- /dev/null
+++ b/klm/util/stream/sort_test.cc
@@ -0,0 +1,62 @@
+#include "util/stream/sort.hh"
+
+#define BOOST_TEST_MODULE SortTest
+#include <boost/test/unit_test.hpp>
+
+#include <algorithm>
+
+#include <unistd.h>
+
+namespace util { namespace stream { namespace {
+
+struct CompareUInt64 : public std::binary_function<const void *, const void *, bool> {
+  bool operator()(const void *first, const void *second) const {
+    return *static_cast<const uint64_t*>(first) < *reinterpret_cast<const uint64_t*>(second);
+  }
+};
+
+const uint64_t kSize = 100000;
+
+struct Putter {
+  Putter(std::vector<uint64_t> &shuffled) : shuffled_(shuffled) {}
+
+  void Run(const ChainPosition &position) {
+    Stream put_shuffled(position);
+    for (uint64_t i = 0; i < shuffled_.size(); ++i, ++put_shuffled) {
+      *static_cast<uint64_t*>(put_shuffled.Get()) = shuffled_[i];
+    }
+    put_shuffled.Poison();
+  }
+  std::vector<uint64_t> &shuffled_;
+};
+
+BOOST_AUTO_TEST_CASE(FromShuffled) {
+  std::vector<uint64_t> shuffled;
+  shuffled.reserve(kSize);
+  for (uint64_t i = 0; i < kSize; ++i) {
+    shuffled.push_back(i);
+  }
+  std::random_shuffle(shuffled.begin(), shuffled.end());
+  
+  ChainConfig config;
+  config.entry_size = 8;
+  config.total_memory = 800;
+  config.block_count = 3;
+
+  SortConfig merge_config;
+  merge_config.temp_prefix = "sort_test_temp";
+  merge_config.buffer_size = 800;
+  merge_config.total_memory = 3300;
+
+  Chain chain(config);
+  chain >> Putter(shuffled);
+  BlockingSort(chain, merge_config, CompareUInt64(), NeverCombine());
+  Stream sorted;
+  chain >> sorted >> kRecycle;
+  for (uint64_t i = 0; i < kSize; ++i, ++sorted) {
+    BOOST_CHECK_EQUAL(i, *static_cast<const uint64_t*>(sorted.Get()));
+  }
+  BOOST_CHECK(!sorted);
+}
+
+}}} // namespaces
-- 
cgit v1.2.3