diff --git a/fsa/fsa_impl.hpp b/fsa/fsa_impl.hpp
index abd6cc6..8d0896b 100644
--- a/fsa/fsa_impl.hpp
+++ b/fsa/fsa_impl.hpp
@@ -18,7 +18,7 @@
#include "utils.hpp"
#include "const.hpp"
-using namespace std;
+//using namespace std;
//static const unsigned int FSA_OFFSET = 6;
template <class T>
@@ -70,7 +70,7 @@ FSA<T>* FSA<T>::getFSA(const unsigned char* ptr, const Deserializer<T>& deserial
uint8_t versionNum = *(ptr + VERSION_NUM_OFFSET);
if (versionNum != VERSION_NUM) {
- throw FSAException(string("Invalid version number: ") + std::to_string(versionNum) + ", should be: " + to_string(VERSION_NUM));
+ throw FSAException(string("Invalid version number: ") + std::to_string(versionNum) + ", should be: " + std::to_string(VERSION_NUM));
}
uint8_t implementationNum = *(ptr + IMPLEMENTATION_NUM_OFFSET);
@@ -85,7 +85,7 @@ FSA<T>* FSA<T>::getFSA(const unsigned char* ptr, const Deserializer<T>& deserial
case 2:
return new CompressedFSA2<T>(startPtr, deserializer);
default:
- throw FSAException(string("Invalid implementation number: ") + to_string(versionNum) + ", should be: " + to_string(VERSION_NUM));
+ throw FSAException(string("Invalid implementation number: ") + std::to_string(versionNum) + ", should be: " + std::to_string(VERSION_NUM));
}
}
diff --git a/morfeusz/CMakeLists.txt b/morfeusz/CMakeLists.txt
index dc0509a..52f710c 100644
--- a/morfeusz/CMakeLists.txt
+++ b/morfeusz/CMakeLists.txt
@@ -8,10 +8,12 @@ add_library (morfeusz2 morfeusz.hpp morfeusz.cpp)
add_executable (morfeusz2_analyze main.cpp)
add_executable (test_morph test_morph.cpp MorphDeserializer.cpp Tagset.cpp ../fsa/const.cpp MorphInterpretation.cpp)
add_executable (test_morfeusz test_morph.cpp MorphDeserializer.cpp Tagset.cpp ../fsa/const.cpp MorphInterpretation.cpp Morfeusz.cpp charset/CharsetConverter.cpp FlexionGraph.cpp)
+add_executable (test_simple test_simple.cpp MorphDeserializer.cpp Tagset.cpp ../fsa/const.cpp MorphInterpretation.cpp Morfeusz.cpp charset/CharsetConverter.cpp FlexionGraph.cpp)
# Link the executable to the Hello library.
target_link_libraries (morfeusz2_analyze morfeusz2)
set_target_properties ( morfeusz2_analyze PROPERTIES COMPILE_FLAGS "-std=gnu++0x" )
set_target_properties ( test_morph PROPERTIES COMPILE_FLAGS "-std=gnu++0x -Wall -O2" )
-set_target_properties ( test_morfeusz PROPERTIES COMPILE_FLAGS "-std=gnu++0x -Wall -O2" )
\ No newline at end of file
+set_target_properties ( test_morfeusz PROPERTIES COMPILE_FLAGS "-std=gnu++0x -Wall -O2" )
+set_target_properties ( test_simple PROPERTIES COMPILE_FLAGS "-std=gnu++0x -Wall -O2" )
\ No newline at end of file
diff --git a/morfeusz/FlexionGraph.cpp b/morfeusz/FlexionGraph.cpp
index 55f5dd0..1aa59be 100644
--- a/morfeusz/FlexionGraph.cpp
+++ b/morfeusz/FlexionGraph.cpp
@@ -30,7 +30,7 @@ void FlexionGraph::addPath(const std::vector<InterpretedChunk>& path) {
void FlexionGraph::appendToResults(const Tagset& tagset, std::vector<MorphInterpretation>& results) {
int endNode = graph.size();
- for (int i = 0; i < graph.size(); i++) {
+ for (unsigned int i = 0; i < graph.size(); i++) {
vector<Edge>& edges = graph[i];
for (Edge& e: edges) {
int realStartNode = i + this->startNode;
diff --git a/morfeusz/InterpretedChunk.hpp b/morfeusz/InterpretedChunk.hpp
index 6e8fd78..c4b74ba 100644
--- a/morfeusz/InterpretedChunk.hpp
+++ b/morfeusz/InterpretedChunk.hpp
@@ -13,7 +13,7 @@
struct InterpretedChunk {
const char* chunk;
long chunkLength;
- InterpsGroup& interpsGroup;
+ InterpsGroup interpsGroup;
};
#endif /* INTERPRETEDCHUNK_HPP */
diff --git a/morfeusz/Morfeusz.cpp b/morfeusz/Morfeusz.cpp
index ef21bf2..99ca9ac 100644
--- a/morfeusz/Morfeusz.cpp
+++ b/morfeusz/Morfeusz.cpp
@@ -6,6 +6,7 @@
*/
#include <string>
+#include <iostream>
#include "fsa.hpp"
#include "utils.hpp"
#include "Morfeusz.hpp"
@@ -18,18 +19,28 @@
using namespace std;
static FSA<vector<InterpsGroup >> *initializeFSA(const string& filename) {
+ cerr << "initialize FSA" << endl;
static Deserializer < vector < InterpsGroup >> *deserializer
= new MorphDeserializer();
return FSA < vector < InterpsGroup >> ::getFSA(filename, *deserializer);
}
static CharsetConverter* initializeCharsetConverter() {
+ cerr << "initialize charset converter" << endl;
static CharsetConverter* converter = new UTF8CharsetConverter();
return converter;
}
+static Tagset* initializeTagset(const string& filename) {
+ cerr << "initialize tagset" << endl;
+ static Tagset* tagset = new Tagset(readFile(filename.c_str()));
+ return tagset;
+}
+
Morfeusz::Morfeusz(const string& filename)
-: fsa(initializeFSA(filename)), charsetConverter(initializeCharsetConverter()) {
+: fsa(initializeFSA(filename)),
+ charsetConverter(initializeCharsetConverter()),
+ tagset(initializeTagset(filename)) {
}
@@ -47,7 +58,7 @@ void Morfeusz::processOneWord(
FlexionGraph graph(startNodeNum);
const char* currInput = inputData;
doProcessOneWord(currInput, inputEnd, accum, graph);
- graph.appendToResults(this->tagset, results);
+ graph.appendToResults(*this->tagset, results);
inputData = currInput;
}
@@ -88,12 +99,16 @@ void Morfeusz::feedState(
}
}
-ResultsIterator Morfeusz::analyze(const std::string& text) {
+ResultsIterator Morfeusz::analyze(const string& text) {
// const char* textStart = text.c_str();
// const char* textEnd = text.c_str() + text.length();
return ResultsIterator(text, *this);
}
+void Morfeusz::analyze(const string& text, vector<MorphInterpretation>& results) {
+
+}
+
ResultsIterator::ResultsIterator(const string& text, const Morfeusz& morfeusz)
: rawInput(text.c_str()),
morfeusz(morfeusz) {
diff --git a/morfeusz/Morfeusz.hpp b/morfeusz/Morfeusz.hpp
index 739f63f..b7a82a8 100644
--- a/morfeusz/Morfeusz.hpp
+++ b/morfeusz/Morfeusz.hpp
@@ -32,15 +32,17 @@ public:
virtual ~Morfeusz();
// Morfeusz(const Morfeusz& orig);
ResultsIterator analyze(const std::string& text);
-
-// Morfeusz();
- friend class ResultsIterator;
-private:
+ void analyze(const std::string& text, std::vector<MorphInterpretation>& result);
+
void processOneWord(
const char*& inputData,
const char* inputEnd,
const int startNodeNum,
std::vector<MorphInterpretation>& result) const;
+
+// Morfeusz();
+ friend class ResultsIterator;
+private:
void doProcessOneWord(
const char*& inputData,
@@ -54,6 +56,7 @@ private:
FSAType* fsa;
CharsetConverter* charsetConverter;
+ Tagset* tagset;
};
class ResultsIterator {
diff --git a/morfeusz/test_simple.cpp b/morfeusz/test_simple.cpp
new file mode 100644
index 0000000..b8afe79
--- /dev/null
+++ b/morfeusz/test_simple.cpp
@@ -0,0 +1,29 @@
+/*
+ * File: test_simple.cpp
+ * Author: lennyn
+ *
+ * Created on November 18, 2013, 10:30 PM
+ */
+
+#include <cstdlib>
+
+#include "Morfeusz.hpp"
+#include "MorphInterpretation.hpp"
+
+using namespace std;
+
+/*
+ *
+ */
+int main(int argc, char** argv) {
+ Morfeusz morfeusz(argv[1]);
+ vector<MorphInterpretation> res;
+ string word = "mijałem";
+ const char* ptr = word.c_str();
+ morfeusz.processOneWord(ptr, word.c_str() + word.size(), 0, res);
+ for (MorphInterpretation& mi: res) {
+ cerr << mi.getStartNode() << " " << mi.getEndNode() << " " << mi.getLemma() << " " << mi.getTag() << " " << mi.getName() << endl;
+ }
+ return 0;
+}
+
diff --git a/nbproject/configurations.xml b/nbproject/configurations.xml
index c0d81ea..60de1ce 100644
--- a/nbproject/configurations.xml
+++ b/nbproject/configurations.xml
@@ -11,12 +11,6 @@
<df name="charset">
<in>CharsetConverter.cpp</in>
</df>
- <df name="encoding">
- <in>CharsetConverter.cpp</in>
- </df>
- <df name="flexion">
- <in>FlexionGraph.cpp</in>
- </df>
<in>FlexionGraph.cpp</in>
<in>Morfeusz.cpp</in>
<in>MorphDeserializer.cpp</in>
@@ -24,6 +18,7 @@
<in>Tagset.cpp</in>
<in>main.cpp</in>
<in>morfeusz.cpp</in>
+ <in>test_morfeusz.cpp</in>
<in>test_morph.cpp</in>
</df>
<logicalFolder name="ExternalFiles"
@@ -57,7 +52,7 @@
<executablePath>build/fsa/test_dict</executablePath>
</makeTool>
</makefileType>
- <folder path="1/charset">
+ <folder path="1">
<ccTool>
<incDir>
<pElem>fsa</pElem>
@@ -96,42 +91,22 @@
</item>
<item path="morfeusz/FlexionGraph.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
<item path="morfeusz/Morfeusz.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
<item path="morfeusz/MorphDeserializer.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
<item path="morfeusz/MorphInterpretation.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
<item path="morfeusz/Tagset.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
<item path="morfeusz/charset/CharsetConverter.cpp"
@@ -141,33 +116,23 @@
<ccTool>
</ccTool>
</item>
- <item path="morfeusz/encoding/CharsetConverter.cpp"
- ex="false"
- tool="1"
- flavor2="4">
+ <item path="morfeusz/main.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
</ccTool>
</item>
- <item path="morfeusz/flexion/FlexionGraph.cpp" ex="false" tool="1" flavor2="4">
+ <item path="morfeusz/morfeusz.cpp" ex="false" tool="1" flavor2="4">
<ccTool>
</ccTool>
</item>
- <item path="morfeusz/main.cpp" ex="false" tool="1" flavor2="8">
+ <item path="morfeusz/test_morfeusz.cpp" ex="false" tool="1" flavor2="4">
<ccTool>
- <incDir>
- <pElem>fsa</pElem>
- <pElem>build/morfeusz</pElem>
- </incDir>
</ccTool>
</item>
- <item path="morfeusz/morfeusz.cpp" ex="false" tool="1" flavor2="4">
+ <item path="morfeusz/test_morph.cpp" ex="false" tool="1" flavor2="8">
<ccTool>
- <incDir>
- <pElem>morfeusz</pElem>
- </incDir>
</ccTool>
</item>
- <item path="morfeusz/test_morph.cpp" ex="false" tool="1" flavor2="8">
+ <item path="morfeusz/test_simple.cpp" ex="false" tool="1" flavor2="0">
<ccTool>
<incDir>
<pElem>fsa</pElem>