nowsza wersja automatu (bez bitu mówiącego o tym, czy trzymać stan w tablicy)

git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/morfeusz@207 ff4e3ee1-f430-4e82-ade0-24591c43f1fd

nowsza wersja automatu (bez bitu mówiącego o tym, czy trzymać stan w tablicy)
git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/morfeusz@207 ff4e3ee1-f430-4e82-ade0-24591c43f1fd
Michał Lenart
1 parent c4676798
Showing 8 changed files with 34 additions and 24 deletions
fsabuilder/morfeusz_builder
fsabuilder/morfeuszbuilder/fsa/serializer.py
morfeusz/Morfeusz.cpp
morfeusz/charset/TextReader.cpp
morfeusz/charset/TextReader.hpp
morfeusz/fsa/cfsa1_impl.hpp
morfeusz/fsa/const.cpp
nbproject/configurations.xml
@@ -258,9 +258,9 @@ if __name__ == &#39;__main__&#39;:
     opts = _parseOptions()
     try:
         main(opts)
-#     except Exception as ex:
-#         print >> sys.stderr, unicode(ex).encode('utf8')
-#         sys.exit(1)
+    except Exception as ex:
+        print >> sys.stderr, unicode(ex).encode('utf8')
+        sys.exit(1)
     finally:
         pass
  
@@ -42,7 +42,7 @@ class Serializer(object):
  
     # get the Morfeusz file format version that is being encoded
     def getVersion(self):
-        return 17
+        return 18
  
     def serialize2CppFile(self, fname, isGenerator, headerFilename="data/default_fsa.hpp"):
         res = []
@@ -236,7 +236,7 @@ class VLengthSerializer1(Serializer):
         self.label2ShortLabel = None
  
         self.ACCEPTING_FLAG = 0b10000000
-        self.ARRAY_FLAG = 0b01000000
+#         self.ARRAY_FLAG = 0b01000000
  
     def getImplementationCode(self):
         return 1
@@ -277,13 +277,13 @@ class VLengthSerializer1(Serializer):
         firstByte = 0
         if state.isAccepting():
             firstByte |= self.ACCEPTING_FLAG
-        if self.stateShouldBeAnArray(state):
-            firstByte |= self.ARRAY_FLAG
-        if state.transitionsNum < 63:
+#         if self.stateShouldBeAnArray(state):
+#             firstByte |= self.ARRAY_FLAG
+        if state.transitionsNum < 127:
             firstByte |= state.transitionsNum
             res.append(firstByte)
         else:
-            firstByte |= 63
+            firstByte |= 127
             res.append(firstByte)
             res.append(state.transitionsNum)
  
@@ -223,7 +223,6 @@ void Morfeusz::doProcessOneWord(
                         }
                         else {
                             assert(!newSegrulesState.sink);
-                            //                        cerr << "will process " << currInput << endl;
                             TextReader newReader(reader.getCurrPtr(), reader.getEndPtr(), env);
                             doProcessOneWord(env, newReader, newSegrulesState);
                         }
 /* 
- * File:   StatefulCharsetConverter.cpp
+ * File:   TextReader.cpp
  * Author: lennyn
  * 
  * Created on May 28, 2014, 11:43 AM
 /* 
- * File:   StatefulCharsetConverter.hpp
+ * File:   TextReader.hpp
  * Author: lennyn
  *
  * Created on May 28, 2014, 11:43 AM
@@ -15,8 +15,8 @@
 #include "../deserializationUtils.hpp"
  
 static const unsigned char CFSA1_ACCEPTING_FLAG = 128;
-static const unsigned char CFSA1_ARRAY_FLAG = 64;
-static const unsigned char CFSA1_TRANSITIONS_NUM_MASK = 63;
+//static const unsigned char CFSA1_ARRAY_FLAG = 64;
+static const unsigned char CFSA1_TRANSITIONS_NUM_MASK = 127;
  
 static const unsigned char CFSA1_OFFSET_SIZE_MASK = 3;
  
@@ -136,14 +136,17 @@ void CompressedFSA1&lt;T&gt;::doProceedToNext(const char c, State&lt;T&gt;&amp; state, bool init
                 if (label == c) {
                     found = true;
                     break;
-                } else {
+                }
+                else {
                     currPtr += td.offsetSize;
                 }
-            } else {
+            } 
+            else {
                 found = true;
                 break;
             }
-        } else {
+        }
+        else {
             if (td.shortLabel == 0) {
                 currPtr++;
             }
@@ -152,7 +155,8 @@ void CompressedFSA1&lt;T&gt;::doProceedToNext(const char c, State&lt;T&gt;&amp; state, bool init
     }
     if (!found) {
         state.setNextAsSink();
-    } else {
+    } 
+    else {
         uint32_t offset;
         switch (td.offsetSize) {
             case 0:
@@ -2,7 +2,7 @@
 #include "const.hpp"
  
 extern const uint32_t MAGIC_NUMBER = 0x8fc2bc1b;
-extern const uint8_t VERSION_NUM = 17;
+extern const uint8_t VERSION_NUM = 18;
  
 extern const unsigned int VERSION_NUM_OFFSET = 4;
 extern const unsigned int IMPLEMENTATION_NUM_OFFSET = 5;
@@ -103,7 +103,8 @@
       <flagsDictionary>
         <element flagsID="0" commonFlags="-O2 -std=c++98"/>
         <element flagsID="1" commonFlags="-O2 -std=c++98 -fPIC"/>
-        <element flagsID="2" commonFlags="-std=c++98 -O3 -fPIC"/>
+        <element flagsID="2" commonFlags="-std=c++98 -O3"/>
+        <element flagsID="3" commonFlags="-std=c++98 -O3 -fPIC"/>
       </flagsDictionary>
       <codeAssistance>
       </codeAssistance>
@@ -145,7 +146,7 @@
         </ccTool>
       </item>
       <item path="build/morfeusz/java/swigJAVA.cpp" ex="false" tool="1" flavor2="4">
-        <ccTool flags="2">
+        <ccTool flags="3">
         </ccTool>
       </item>
       <item path="build/morfeusz/morfeuszJAVA_wrap.cxx"
@@ -382,14 +383,15 @@
         </ccTool>
       </item>
       <item path="morfeusz/Morfeusz.cpp" ex="false" tool="1" flavor2="4">
-        <ccTool flags="1">
+        <ccTool flags="3">
           <preprocessorList>
+            <Elem>NDEBUG</Elem>
             <Elem>libmorfeusz_EXPORTS</Elem>
           </preprocessorList>
         </ccTool>
       </item>
       <item path="morfeusz/MorphDeserializer.cpp" ex="false" tool="1" flavor2="4">
-        <ccTool flags="2">
+        <ccTool flags="3">
           <preprocessorList>
             <Elem>NDEBUG</Elem>
             <Elem>libmorfeusz_EXPORTS</Elem>
@@ -425,7 +427,7 @@
             ex="false"
             tool="1"
             flavor2="4">
-        <ccTool flags="2">
+        <ccTool flags="3">
           <preprocessorList>
             <Elem>NDEBUG</Elem>
           </preprocessorList>
@@ -532,6 +534,11 @@
         </ccTool>
       </item>
       <item path="morfeusz/morfeusz_analyzer.cpp" ex="false" tool="1" flavor2="4">
+        <ccTool flags="2">
+          <preprocessorList>
+            <Elem>NDEBUG</Elem>
+          </preprocessorList>
+        </ccTool>
       </item>
       <item path="morfeusz/morfeusz_generator.cpp" ex="false" tool="1" flavor2="4">
       </item>