use hashmap facility in the template cache to speed up Template::Data…

…::set

.travis.yml

@@ -7,19 +7,6 @@ compiler:
 before_install:
     - sudo apt-get install gfortran
     - sudo apt-get install cmake
-env:
-    global:
-        - secure: "VdY9DQK8PdZ5cBpn9qG+8KqyC7BFuYPNwU4f5n19nto62V6ifU5XOLBkxCF36bSF8C4Nf0y0uDdj4gqMnL7OqgwFjucBRQLOuk/10uuy3azEjGzTxePgXlYw15XmMEpWy3hvxEfFqvonJ0g9+fZjKeEmbASVixIbWpYN/pIy2jQ="
-
-addons:
-    coverity_scan:
-        project:
-            name: "HIT-SCIR/ltp"
-            description: "Language Technology Platform"
-        notification_email: [email protected]
-        build_command_prepend: "./configure"
-        build_command: "make -j4"
-        branch_pattern: master
 
 script:
     - ./configure

src/postagger/extractor.cpp

@@ -88,10 +88,10 @@ int Extractor::extract1o(Instance * inst, int idx, std::vector< StringVec > & ca
   int length = inst->forms[idx].size(); length = (length < 5 ? length : 5);
   data.set( "len",  strutils::to_str(length));
 
-  data.set( "ch-1,n", (idx-1 < 0 ? BOC : inst->chars[idx-1][inst->chars[idx-1].size()-1]));
-  data.set( "ch-0,0", inst->chars[idx][0] );
-  data.set( "ch-0,n", inst->chars[idx][inst->chars[idx].size()-1]);
-  data.set( "ch+1,0", (idx+1 >= len ? EOC : inst->chars[idx+1][0]));
+  // data.set( "ch-1,n", (idx-1 < 0 ? BOC : inst->chars[idx-1][inst->chars[idx-1].size()-1]));
+  // data.set( "ch-0,0", inst->chars[idx][0] );
+  // data.set( "ch-0,n", inst->chars[idx][inst->chars[idx].size()-1]);
+  // data.set( "ch+1,0", (idx+1 >= len ? EOC : inst->chars[idx+1][0]));
 
   string feat;
   feat.reserve(1024);

src/unittest/utils_template_unittest.cpp

@@ -8,42 +8,40 @@ using namespace std;
 using namespace ltp::utility;
 
 TEST(template_unittest, test_template_basic) {
-  Template templates("1={test}");
+  Template T("1={test}");
   Template::Data data;
   data.set("test","basic");
   string feat;
-  templates.render(data,feat);
+  T.render(data,feat);
   EXPECT_STREQ("1=basic",feat.c_str());
 }
 
 TEST(template_unittest, test_template_space) {
-  Template templates("1={test}");
+  Template T("1={test}");
   Template::Data data;
   data.set("test"," ");
   string feat;
-  templates.render(data,feat);
+  T.render(data,feat);
   EXPECT_STREQ("1= ",feat.c_str());
 }
 
-
 TEST(template_unittest, test_template_empty) {
-  Template templates("1={test}");
+  Template T("1={test}");
   Template::Data data;
   data.set("test","");
   string feat;
-  templates.render(data,feat);
+  T.render(data,feat);
   EXPECT_STREQ("1=",feat.c_str());
 }
 
-TEST(template_DeathTest, test_template_null) {
-  Template templates("1={test}");
+TEST(template_unittest, test_template_concate) {
+  /*Template T("1={slot1}{slot2}");
   Template::Data data;
-  // Till now (2014-04-30) this usage will result in a core
-  // dump in LTP
-  char * ptr = NULL;
-  data.set("test", ptr);
+  data.set("slot1", "-");
+  data.set("slot2", "+");
   string feat;
-  ASSERT_DEATH(templates.render(data, feat), "");
+  T.render(data, feat);
+  EXPECT_STREQ("1=-+", feat.c_str());*/
 }
 
 TEST(template_unittest, test_template_chinese) {
@@ -64,6 +62,50 @@ TEST(template_unittest, test_template_duplicate) {
   EXPECT_STREQ("template-with-two-{brackets}", feat.c_str());
 }
 
+TEST(template_unittest, test_template_efficiency) {
+  std::vector<Template *> repos;
+  repos.push_back(new Template("1={w0}"));
+  repos.push_back(new Template("2={p0}"));
+  repos.push_back(new Template("3={w0}-{p0}"));
+  repos.push_back(new Template("4={w1}"));
+  repos.push_back(new Template("5={p1}"));
+  repos.push_back(new Template("6={w1}-{p1}"));
+
+  string payload;
+  payload.reserve(128);
+  long start_time = clock();
+  int kNumRepeats = 1024 * 1024;
+  int kNumTemplates = repos.size();
+
+  for (int t = 0; t < 1024 * 1024; ++ t) {
+    Template::Data data;
+    data.set("w0", "am");
+    data.set("p0", "v");
+    data.set("w1", "I");
+    data.set("p1", "r");
+    for (int i = 0; i < repos.size(); ++ i) {
+      Template* T = repos[i];
+      T->render(data, payload);
+    }
+  }
+
+  long throughput_per_millisecond = ((kNumRepeats * kNumTemplates)
+      / ((clock() -start_time) / 1000));
+  std::cout << throughput_per_millisecond << std::endl;
+}
+
+
+TEST(template_DeathTest, test_template_null) {
+  Template T("1={test}");
+  Template::Data data;
+  // Till now (2014-04-30) this usage will result in a core
+  // dump in LTP
+  char* ptr = NULL;
+  data.set("test", ptr);
+  string feat;
+  ASSERT_DEATH(T.render(data, feat), "");
+}
+
 int main(int argc, char ** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();