nlp work

djmott · Aug 16, 2016 · 7b9bed8 · 7b9bed8
1 parent 78445f0
commit 7b9bed8
Show file tree

Hide file tree

Showing 5 changed files with 32 additions and 20 deletions.
diff --git a/examples/example_nlp.cpp b/examples/example_nlp.cpp
@@ -19,6 +19,7 @@ namespace commands{
 
 xtd::nlp::english::pointer _english;
 
+#if 0
 int main(){
   using namespace xtd;
 
@@ -42,9 +43,11 @@ int main(){
   return 0;
 
 }
-#if 0
+#else
 int main(){
-  _english = xtd::nlp::english::pointer(new xtd::nlp::english("/home/davidmott/Downloads/Moby", "/home/davidmott/Downloads/WordNet-3.0/dict"));
+
+  xtd::nlp::wordnet::database oDB;
+
 
   std::cout << "Chatty Cathy: Hello! What can I do for you?" << std::endl;
 

diff --git a/include/xtd/nlp/english.hpp b/include/xtd/nlp/english.hpp
@@ -30,7 +30,7 @@ namespace xtd{
       const lemma::lemmata& lemmata() const { return _lemmata; }
 
 
-      english(const xtd::filesystem::path& oMobyPath, const xtd::filesystem::path& oWordnetPath) : _moby(oMobyPath), _wordnet(oWordnetPath){
+      english() : _wordnet(){
         for (const auto & oRecord : _wordnet._data_adj.records){
           for (const auto & oWord : oRecord.second.words){
             std::cout << oWord.word << std::endl;
@@ -40,7 +40,6 @@ namespace xtd{
       }
 
     private:
-      moby::database _moby;
       wordnet::database _wordnet;
       lemma::lemmata _lemmata;
     };

diff --git a/include/xtd/nlp/nlp.hpp b/include/xtd/nlp/nlp.hpp
@@ -82,13 +82,10 @@ namespace xtd{
         using type = _Ty;
       };
 
+    }
 
 
 
-
-
-    }
-
     struct BOW{
       std::map<std::string, size_t> operator()(const std::vector<xtd::string>& oDoc) const{
         std::map<std::string, size_t> oRet;
@@ -99,13 +96,15 @@ namespace xtd{
       }
     };
 
+
     struct SimpleParse{
       std::vector<xtd::string> operator()(const std::string& src) const {
         return xtd::string(src).split({'.', '!', '?', ' ', ','});
       }
     };
 
 
+
   }
 }
 

diff --git a/include/xtd/nlp/wordnet.hpp b/include/xtd/nlp/wordnet.hpp
@@ -8,6 +8,7 @@ c++ interface to wordnet databases
 
 #include <map>
 #include <future>
+#include <sstream>
 
 #include <xtd/filesystem.hpp>
 
@@ -52,8 +53,13 @@ namespace xtd
           using map = std::map<size_t, record>;
           size_t file_offset;
           std::string lemma, pos, synset_cnt, p_cnt, ptr_symbol, sense_cnt, tagsense_cnt, synset_offset;
-          bool load(const std::string& , size_t & ) {
-            return false;
+          bool load(const std::string& str, size_t & i) {
+            auto x = str.find('\n', i);
+            std::stringstream oSS;
+            oSS.str(std::string(&str[i], &str[x]));
+            oSS >> lemma >> pos >> synset_cnt >> p_cnt >> ptr_symbol >> sense_cnt >> tagsense_cnt >> synset_offset;
+            i = x;
+            return true;
           }
         };
 
@@ -202,35 +208,36 @@ namespace xtd
         index_file _index_noun;
         index_file _index_verb;
 
-        database(const xtd::filesystem::path& oPath) {
+        database() {
           auto make_path = [&](const char * sAddend){
-            xtd::filesystem::path oRet(oPath);
+            xtd::filesystem::path oRet(XTD_ASSETS_DIR "/WordNet-3.0/dict");
             oRet /= sAddend;
+            oRet.make_preferred();
             return oRet;
           };
           auto t1 = std::async(std::launch::async, [&]() {
-            return make_path("data.adj");
+            return _data_adj.load(make_path("data.adj"));
           });
           auto t2 = std::async(std::launch::async, [&]() {
-            return make_path("data.adv");
+            return _data_adv.load(make_path("data.adv"));
           });
           auto t3 = std::async(std::launch::async, [&]() {
-            return make_path("data.noun");
+            return _data_noun.load(make_path("data.noun"));
           });
           auto t4 = std::async(std::launch::async, [&]() {
-            return make_path("data.verb");
+            return _data_verb.load(make_path("data.verb"));
           });
           auto t5 = std::async(std::launch::async, [&]() {
-            return make_path("index.adj");
+            return _index_adj.load(make_path("index.adj"));
           });
           auto t6 = std::async(std::launch::async, [&]() {
-            return make_path("index.adv");
+            return _index_adv.load(make_path("index.adv"));
           });
           auto t7 = std::async(std::launch::async, [&]() {
-            return make_path("index.noun");
+            return _index_noun.load(make_path("index.noun"));
           });
           auto t8 = std::async(std::launch::async, [&]() {
-            return make_path("index.verb");
+            return _index_verb.load(make_path("index.verb"));
           });
           t1.get();
           t2.get();

diff --git a/include/xtd/xtd.hpp.in b/include/xtd/xtd.hpp.in
@@ -28,6 +28,10 @@
   #define DOXY_INVOKED 0
 #endif
 
+#if !defined(XTD_ASSETS_DIR)
+  #define XTD_ASSETS_DIR "@CMAKE_CURRENT_SOURCE_DIR@/assets"
+#endif
+
 #if !defined(XTD_HAS_LIBUUID)
   #cmakedefine01 XTD_HAS_LIBUUID
 #endif