- dalsza praca nad kompilacją do Javy

git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/morfeusz@52 ff4e3ee1-f430-4e82-ade0-24591c43f1fd

- dalsza praca nad kompilacją do Javy
git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/morfeusz@52 ff4e3ee1-f430-4e82-ade0-24591c43f1fd
Michał Lenart
1 parent a7ab50de
Showing 7 changed files with 75 additions and 63 deletions
CMakeLists.txt
morfeusz/Morfeusz.cpp
morfeusz/Morfeusz.hpp
morfeusz/MorphInterpretation.hpp
morfeusz/java/CMakeLists.txt
morfeusz/morfeusz.i
nbproject/configurations.xml
@@ -16,9 +16,22 @@ configure_file (
  
 include_directories("${PROJECT_BINARY_DIR}" )
  
+set (CMAKE_LIBRARY_OUTPUT_DIRECTORY "${PROJECT_BINARY_DIR}/lib")
 add_subdirectory (morfeusz)
  
-file(COPY fsabuilder testfiles DESTINATION .)
+# copy jmorfeusz to java source tree
+file(GLOB files "${CMAKE_LIBRARY_OUTPUT_DIRECTORY}/*jmorfeusz*")
+foreach(file ${files})
+  file (COPY ${file} DESTINATION ${PROJECT_SOURCE_DIR}/jmorfeusz/src/main/native)
+endforeach()
+
+# build jmorfeusz
+add_custom_command(TARGET all POST_BUILD mvn package -PotherOutputDir -DoutputDir=${PROJECT_BINARY_DIR}/jmorfeusz/target WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}/jmorfeusz)
+
+
+# TESTS
+
+file (COPY fsabuilder testfiles DESTINATION .)
  
 macro (test_build_and_recognize fname method)
     add_test (TestBuild-${method}-${fname} python fsabuilder/fsa/buildfsa.py -i testfiles/${fname} -o /tmp/test-${method}-${fname}.fsa --tagset-file=testfiles/polimorf.tagset --output-format=BINARY --serialization-method=${method})
@@ -226,18 +226,17 @@ void Morfeusz::setEncoding(MorfeuszCharset encoding) {
 ResultsIterator::ResultsIterator(const string& text, const Morfeusz& morfeusz)
 : rawInput(text.c_str()),
 morfeusz(morfeusz) {
+    vector<MorphInterpretation> res;
+    morfeusz.analyze(text, res);
+    resultsBuffer.insert(resultsBuffer.begin(), res.begin(), res.end());
 }
  
 MorphInterpretation ResultsIterator::getNext() {
-    //    if (resultsBuffer.empty()) {
-    //        morfeusz.processOneWord(rawInput, startNode, back_inserter(resultsBuffer));
-    //    }
-    //    startNode = resultsBuffer.back().getEndNode();
-    //    MorphInterpretation res = resultsBuffer.front();
-    //    resultsBuffer.pop_front();
-    //    return res;
+    MorphInterpretation res = this->resultsBuffer.front();
+    this->resultsBuffer.pop_front();
+    return res;
 }
  
 bool ResultsIterator::hasNext() {
-    return rawInput[0] != '\0' && resultsBuffer.empty();
+    return !resultsBuffer.empty();
 }
@@ -22,11 +22,21 @@
 #include "MorfeuszOptions.hpp"
 #include "const.hpp"
  
+class Morfeusz;
+class ResultsIterator;
+
+typedef FSA<std::vector<InterpsGroup > > FSAType;
+typedef State<std::vector<InterpsGroup > > StateType;
+
 class MorfeuszException : public std::exception {
 public:
-    MorfeuszException(const char* what): msg(what) {}
-    MorfeuszException(const std::string& what): msg(what.c_str()) {}
-    virtual ~MorfeuszException() throw() {}
+
+    MorfeuszException(const std::string& what) : msg(what.c_str()) {
+    }
+
+    virtual ~MorfeuszException() throw () {
+    }
+
     virtual const char* what() const throw () {
         return this->msg.c_str();
     }
@@ -34,13 +44,6 @@ private:
     const std::string msg;
 };
  
-class Morfeusz;
-//class AnalyzeResult;
-class ResultsIterator;
-
-typedef FSA<std::vector<InterpsGroup > > FSAType;
-typedef State<std::vector<InterpsGroup > > StateType;
-
 class Morfeusz {
 public:
     Morfeusz();
@@ -50,17 +53,17 @@ public:
     ResultsIterator analyze(const std::string& text) const;
     void analyze(const std::string& text, std::vector<MorphInterpretation>& result) const;
  
+    void setEncoding(MorfeuszCharset encoding);
+
+    //    Morfeusz();
+    friend class ResultsIterator;
+private:
+
     void processOneWord(
             const char*& inputData,
             const char* inputEnd,
             int startNodeNum,
             std::vector<MorphInterpretation>& result) const;
-    
-    void setEncoding(MorfeuszCharset encoding);
-    
-    //    Morfeusz();
-    friend class ResultsIterator;
-private:
  
     void doProcessOneWord(
             const char*& inputData,
@@ -81,48 +84,24 @@ private:
     CharsetConverter* charsetConverter;
     Tagset* tagset;
     CaseConverter* caseConverter;
-    
+
     UTF8CharsetConverter utf8CharsetConverter;
-    
+
     MorfeuszOptions options;
 };
  
 class ResultsIterator {
 public:
-    ResultsIterator(const std::string& text, const Morfeusz& morfeusz);
     MorphInterpretation getNext();
     bool hasNext();
+    friend class Morfeusz;
 private:
+    ResultsIterator(const std::string& text, const Morfeusz& morfeusz);
     const char* rawInput;
     const Morfeusz& morfeusz;
     std::list<MorphInterpretation> resultsBuffer;
     int startNode;
 };
  
-//class ResultsIterator {
-//public:
-//    ResultsIterator(
-//                const char* startOfInput,
-//                const char* endOfInput,
-//                const Morfeusz& morfeusz);
-//    virtual ~ResultsIterator();
-//    ResultsIterator(const ResultsIterator& mit);
-//    ResultsIterator& operator++();
-//    ResultsIterator operator++(int);
-//    bool operator==(const ResultsIterator& rhs);
-//    bool operator!=(const ResultsIterator& rhs);
-//    MorphInterpretation& operator*();
-//private:
-//    const char* rawInput;
-//    const char* endOfInput;
-//    const Morfeusz& morfeusz;
-//    vector<MorphInterpretation> resultsBuffer;
-//};
-
-//struct AnalyzeResult {
-//    ResultsIterator iterator;
-//    const ResultsIterator end;
-//};
-
 #endif	/* MORFEUSZ_HPP */
  
@@ -22,7 +22,6 @@ public:
             const std::string& lemma,
             int tagnum,
             int namenum,
-//            const EncodedInterpretation& encodedInterp,
             const Tagset& tagset,
             const CharsetConverter& charsetConverter);
     static MorphInterpretation createIgn(int startNode, const std::string& orth, const Tagset& tagset, const CharsetConverter& charsetConverter);
@@ -8,15 +8,15 @@ include(UseJava)
  
 # SWIG Java
 INCLUDE_DIRECTORIES(${JAVA_INCLUDE_PATH})
-#INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR})
 INCLUDE_DIRECTORIES(..)
  
-SET(CMAKE_SWIG_FLAGS "")
+set (CMAKE_SWIG_FLAGS -package pl.waw.ipipan.morfeusz)
+set (CMAKE_SWIG_OUTDIR ${CMAKE_SOURCE_DIR}/jmorfeusz/src/main/java/pl/waw/ipipan/morfeusz) 
  
 SET_SOURCE_FILES_PROPERTIES(../morfeusz.i PROPERTIES CPLUSPLUS ON)
-SWIG_ADD_MODULE(morfeusz-java java ../morfeusz.i)
-SWIG_LINK_LIBRARIES(morfeusz-java ${JAVA_LIBRARIES})
-SWIG_LINK_LIBRARIES(morfeusz-java libmorfeusz)
+SWIG_ADD_MODULE(jmorfeusz java ../morfeusz.i)
+SWIG_LINK_LIBRARIES(jmorfeusz ${JAVA_LIBRARIES})
+SWIG_LINK_LIBRARIES(jmorfeusz libmorfeusz)
  
 set (CMAKE_SHARED_LINKER_FLAGS "")
 set (CMAKE_EXE_LINKER_FLAGS "")
@@ -4,23 +4,45 @@
 %{
 #include "Morfeusz.hpp"
 #include "MorphInterpretation.hpp"
+#include "const.hpp"
 %}
  
 %include "std_vector.i"
 %include "std_string.i"
  
 #ifdef SWIGJAVA
+
 %include "enums.swg"
+
 /* Force the generated Java code to use the C enum values rather than making a JNI call */
 %javaconst(1);
+
+%pragma(java) jniclasscode=%{
+  static {
+    System.loadLibrary("jmorfeusz");
+  }
+%}
 #endif
  
-// Very simple C++ example for linked list
+%ignore MorphInterpretation::MorphInterpretation(
+            int startNode,
+            int endNode,
+            const std::string& orth,
+            const std::string& lemma,
+            int tagnum,
+            int namenum,
+            const Tagset& tagset,
+            const CharsetConverter& charsetConverter);
+
+%ignore MorphInterpretation::createIgn(int startNode, const std::string& orth, const Tagset& tagset, const CharsetConverter& charsetConverter);
+
+%ignore Tagset::Tagset(const unsigned char* fsaData);
  
 %include "Morfeusz.hpp"
 %include "MorphInterpretation.hpp"
 %include "const.hpp"
  
+// instantiate vector of interpretations
 namespace std {
        // dirty hack so it will compile without no-arg constructor for MorphInterpretation
    %ignore vector<MorphInterpretation>::vector(size_type); 
@@ -68,7 +68,7 @@
       <item path="build/morfeusz/morfeuszJAVA_wrap.cxx"
             ex="false"
             tool="1"
-            flavor2="4">
+            flavor2="8">
         <ccTool>
           <incDir>
             <pElem>/usr/lib/jvm/default-java/include</pElem>
@@ -76,14 +76,14 @@
             <pElem>build/morfeusz/java</pElem>
           </incDir>
           <preprocessorList>
-            <Elem>morfeusz_java_EXPORTS</Elem>
+            <Elem>jmorfeusz_EXPORTS</Elem>
           </preprocessorList>
         </ccTool>
       </item>
       <item path="build/morfeusz/morfeuszPYTHON_wrap.cxx"
             ex="false"
             tool="1"
-            flavor2="4">
+            flavor2="8">
         <ccTool>
           <incDir>
             <pElem>/usr/include/python2.7</pElem>