SGJP / Morfeusz

Commit a65ee056dbe81292da3af8dcacf7f040494e0d8f

Authored by Michał Lenart
1 parent a1d73c9e

generalne porządki w kodzie 

git-svn-id: svn://svn.nlp.ipipan.waw.pl/morfeusz/morfeusz@210 ff4e3ee1-f430-4e82-ade0-24591c43f1fd
Inline Side-by-side
Showing 71 changed files with 470 additions and 337 deletions
CMakeLists.txt
  View file @a65ee05
... ... @@ -6,7 +6,7 @@ set (Morfeusz_VERSION_MAJOR 2)
6 6 set (Morfeusz_VERSION_MINOR 0)
7 7 set (Morfeusz_VERSION_PATCH 0)
8 8 set (Morfeusz_VERSION "${Morfeusz_VERSION_MAJOR}.${Morfeusz_VERSION_MINOR}.${Morfeusz_VERSION_PATCH}")
9   -if (NOT ${VERSION_SUFFIX} STREQUAL "")
  9 +if (VERSION_SUFFIX)
10 10 set (Morfeusz_VERSION_TWEAK "${VERSION_SUFFIX}")
11 11 set (Morfeusz_VERSION "${Morfeusz_VERSION}_${Morfeusz_VERSION_TWEAK}")
12 12 endif ()
... ... @@ -148,13 +148,11 @@ include (CPack)
148 148 file (COPY fsabuilder testfiles input DESTINATION .)
149 149  
150 150 configure_file (
151   - "${PROJECT_SOURCE_DIR}/morfeusz/MorfeuszConfig.hpp.in"
152   - "${PROJECT_BINARY_DIR}/morfeusz/MorfeuszConfig.hpp"
  151 + "${PROJECT_SOURCE_DIR}/morfeusz/MorfeuszVersion.hpp.in"
  152 + "${PROJECT_BINARY_DIR}/MorfeuszVersion.hpp"
153 153 )
154   -
155   -###### add main sources ########
156   -
157 154 include_directories("${PROJECT_BINARY_DIR}" )
  155 +###### add main sources ########
158 156  
159 157 add_subdirectory (morfeusz)
160 158 add_subdirectory (fsabuilder)
... ...
... ... @@ -109,7 +109,7 @@ Create separate build directory, for example build-darwin.
109 109 Run cross-compilation:
110 110 ----------------------
111 111  
112   -cmake -D CROSSMORFEUSZ_ROOT=<path_to_crossmorfeusz_dir> -DCMAKE_TOOLCHAIN_FILE=../morfeusz/Toolchain-xxx.cmake ..
  112 +cmake -D CROSSMORFEUSZ_ROOT=<path_to_crossmorfeusz_dir> -DCMAKE_TOOLCHAIN_FILE=../toolchains/Toolchain-xxx.cmake ..
113 113 make
114 114 make package package-java package-python
115 115  
... ...
buildAll.sh
  View file @a65ee05
... ... @@ -30,12 +30,12 @@ function build {
30 30 srcDir=`pwd`
31 31 buildDir=buildall/$os-$arch
32 32 targetDir=$srcDir/target
33   - toolchain=$srcDir/morfeusz/Toolchain-$os-$arch.cmake
  33 + toolchain=$srcDir/toolchains/Toolchain-$os-$arch.cmake
34 34  
35 35 echo "Will use $toolchain toolchain"
36 36  
37 37 rm -rf $buildDir
38   - rm -rf $targetDir
  38 + #~ rm -rf $targetDir
39 39 mkdir -p $buildDir
40 40 mkdir -p $targetDir
41 41 cd $buildDir
... ... @@ -68,7 +68,7 @@ function log {
68 68 export -f build
69 69 export -f log
70 70  
71   -rm -rf log
  71 +rm -rf log target
72 72 mkdir -p log
73 73  
74 74 buildDictionaries 2>&1 | log All all
... ... @@ -79,6 +79,6 @@ buildDictionaries 2&gt;&amp;1 | log All all
79 79 echo "build Windows amd64 package package-java 2>&1 | log Windows amd64"
80 80 echo "build Windows i386 package package-java 2>&1 | log Windows i386"
81 81 echo "build Darwin amd64 package package-java 2>&1 | log Darwin amd64"
82   -} | xargs -n1 -P8 -d$'\n' bash -c
  82 +} | xargs -n1 -P5 -d$'\n' bash -c
83 83  
84 84  
... ...
morfeusz/CMakeLists.txt
  View file @a65ee05
... ... @@ -25,8 +25,6 @@ add_custom_target ( dictionaries DEPENDS analyzer-dictionary generator-dictionar
25 25  
26 26 include_directories( ${CMAKE_CURRENT_SOURCE_DIR} )
27 27  
28   -# add_custom_target (dupa DEPENDS "${INPUT_DICTIONARY_CPP}")
29   -
30 28 #### build #####
31 29  
32 30 set(SRC_FILES
... ... @@ -42,37 +40,39 @@ set(SRC_FILES
42 40 InflexionGraph.cpp
43 41 charset/TextReader.cpp
44 42 charset/CharsetConverter.cpp
45   - charset/CaseConverter.cpp
46   - charset/caseconv.cpp
  43 + case/CaseConverter.cpp
  44 + case/caseconv.cpp
47 45 charset/conversion_tables.cpp
48 46 cli/cli.cpp
49 47 segrules/segrules.cpp
50 48 segrules/SegrulesFSA.cpp
51   - CasePatternHelper.cpp
52   - decoder/InterpretedChunksDecoder.cpp
53   - decoder/InterpretedChunksDecoder4Analyzer.cpp
54   - decoder/InterpretedChunksDecoder4Generator.cpp
55   - deserializer/InterpsGroupsReader.cpp
56   - deserializer/MorphDeserializer.cpp
  49 + case/CasePatternHelper.cpp
  50 + deserialization/morphInterps/InterpretedChunksDecoder.cpp
  51 + deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.cpp
  52 + deserialization/morphInterps/InterpretedChunksDecoder4Generator.cpp
  53 + deserialization/InterpsGroupsReader.cpp
  54 + deserialization/MorphDeserializer.cpp
57 55 )
58 56  
59 57 set(INCLUDE_FILES
60 58 const.hpp
61 59 data/default_fsa.hpp
  60 + Environment.hpp
62 61 Tagset.hpp
63 62 Qualifiers.hpp
64 63 fsa/const.hpp
65 64 MorphInterpretation.hpp
66 65 Morfeusz.hpp
  66 + MorfeuszVersion.hpp
67 67 InflexionGraph.hpp
68 68 charset/CharsetConverter.hpp
69 69 charset/TextReader.hpp
70   - charset/CaseConverter.hpp
71   - charset/caseconv.hpp
  70 + case/CaseConverter.hpp
  71 + case/caseconv.hpp
72 72 charset/conversion_tables.hpp
73 73 cli/cli.hpp
74 74 segrules/segrules.hpp
75   - deserializer/MorphDeserializer.cpp
  75 + deserialization/MorphDeserializer.cpp
76 76 )
77 77  
78 78 add_library (libmorfeusz SHARED ${SRC_FILES})
... ... @@ -82,24 +82,15 @@ set_target_properties (libmorfeusz PROPERTIES OUTPUT_NAME &quot;morfeusz2&quot;)
82 82  
83 83 add_executable (morfeusz_analyzer morfeusz_analyzer.cpp)
84 84 add_executable (morfeusz_generator morfeusz_generator.cpp)
85   -add_executable (test_result_equals test_result_equals.cpp)
86   -add_executable (test_recognize_dict test_recognize_dict.cpp)
  85 +add_executable (test_result_equals test/test_result_equals.cpp)
  86 +add_executable (test_recognize_dict test/test_recognize_dict.cpp)
87 87  
88 88 target_link_libraries (morfeusz_analyzer libmorfeusz)
89 89 target_link_libraries (morfeusz_generator libmorfeusz)
90 90 target_link_libraries (test_result_equals libmorfeusz)
91 91 target_link_libraries (test_recognize_dict libmorfeusz)
92 92  
93   -if (${CMAKE_SYSTEM_NAME} MATCHES "Windows")
94   - target_link_libraries (libmorfeusz ws2_32)
95   - set (TARGET_LIB_DIR bin)
96   -else ()
97   - set (TARGET_LIB_DIR lib)
98   -endif ()
99   -
100   -add_subdirectory (java)
101   -add_subdirectory (python)
102   -add_subdirectory (perl)
  93 +add_subdirectory (wrappers)
103 94  
104 95 if (${CMAKE_SYSTEM_NAME} MATCHES "Darwin")
105 96 add_custom_target (morfeusz-repair-library
... ... @@ -108,6 +99,13 @@ if (${CMAKE_SYSTEM_NAME} MATCHES &quot;Darwin&quot;)
108 99 add_dependencies (morfeusz_analyzer morfeusz-repair-library)
109 100 endif ()
110 101  
  102 +if (${CMAKE_SYSTEM_NAME} MATCHES "Windows")
  103 + target_link_libraries (libmorfeusz ws2_32)
  104 + set (TARGET_LIB_DIR bin)
  105 +else ()
  106 + set (TARGET_LIB_DIR lib)
  107 +endif ()
  108 +
111 109 install (FILES ${INCLUDE_FILES} DESTINATION include/morfeusz)
112 110 install (TARGETS libmorfeusz DESTINATION ${TARGET_LIB_DIR})
113 111 install (TARGETS morfeusz_analyzer morfeusz_generator DESTINATION bin)
... ...
morfeusz/Environment.cpp
  View file @a65ee05
... ... @@ -8,14 +8,11 @@
8 8 #include <vector>
9 9 #include <algorithm>
10 10 #include "Environment.hpp"
11   -#include "decoder/InterpretedChunksDecoder.hpp"
12   -#include "deserializer/MorphDeserializer.hpp"
  11 +#include "deserialization/MorphDeserializer.hpp"
13 12 #include "exceptions.hpp"
14   -#include "decoder/InterpretedChunksDecoder4Analyzer.hpp"
15   -#include "decoder/InterpretedChunksDecoder4Generator.hpp"
16   -
17   -//class InterpretedChunksDecoder4Analyzer;
18   -//class InterpretedChunksDecoder4Generator;
  13 +#include "deserialization/morphInterps/InterpretedChunksDecoder.hpp"
  14 +#include "deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.hpp"
  15 +#include "deserialization/morphInterps/InterpretedChunksDecoder4Generator.hpp"
19 16  
20 17 static Deserializer<InterpsGroupsReader>& initializeDeserializer(MorfeuszProcessorType processorType) {
21 18 static Deserializer<InterpsGroupsReader> *analyzerDeserializer
... ...
morfeusz/Environment.hpp
  View file @a65ee05
... ... @@ -13,56 +13,142 @@
13 13 class InterpretedChunksDecoder;
14 14 class CasePatternHelper;
15 15  
16   -#include "charset/CaseConverter.hpp"
  16 +#include "case/CaseConverter.hpp"
17 17 #include "charset/CharsetConverter.hpp"
18 18 #include "fsa/fsa.hpp"
19 19 #include "segrules/segrules.hpp"
20 20 #include "const.hpp"
21 21 #include "Tagset.hpp"
22   -//#include "InterpretedChunksDecoder.hpp"
23 22 #include "InterpsGroup.hpp"
24   -#include "CasePatternHelper.hpp"
  23 +#include "case/CasePatternHelper.hpp"
25 24 #include "Qualifiers.hpp"
26   -#include "deserializer/InterpsGroupsReader.hpp"
  25 +#include "deserialization/InterpsGroupsReader.hpp"
27 26  
28 27 struct InterpsGroup;
29 28 typedef FSA<InterpsGroupsReader> FSAType;
30   -//typedef FSA< std::vector<InterpsGroup > > FSAType;
31 29  
  30 +/**
  31 + * This class contains data required for morphological analysis/synthesis.
  32 + * It contains references to dictionary automaton, charset converter, tagset data etc.
  33 + * All of these can be changed by setters, changing Morfeusz behavior (different dictionary, charset, and other options).
  34 + */
32 35 class Environment {
33 36 public:
  37 + /**
  38 + * Creates default environment with given initial charset, processor type (analyzer/generator) and default dictionary data ptr.
  39 + *
  40 + * @param charset
  41 + * @param morfeuszProcessor
  42 + * @param fileStartPtr
  43 + */
34 44 Environment(
35 45 MorfeuszCharset charset,
36 46 MorfeuszProcessorType morfeuszProcessor,
37 47 const unsigned char* fileStartPtr);
38 48  
  49 + /**
  50 + * Sets charset for this environment.
  51 + *
  52 + * @param charset
  53 + */
39 54 void setCharset(MorfeuszCharset charset);
40 55  
  56 + /**
  57 + * Sets case sensitivity options.
  58 + *
  59 + * @param caseSensitive - if true, interpretations not matching case will be discarded.
  60 + */
41 61 void setCaseSensitive(bool caseSensitive);
42 62  
  63 + /**
  64 + * Gets charset converter that is currently used by this environment.
  65 + * Changed by setting charset.
  66 + *
  67 + * @return - reference to charset converter.
  68 + */
43 69 const CharsetConverter& getCharsetConverter() const;
44 70  
  71 + /**
  72 + * Returns case converter that is currently used by this environment.
  73 + * Changed by setting case sensitivity option.
  74 + *
  75 + * @return - reference to case converter.
  76 + */
45 77 const CaseConverter& getCaseConverter() const;
46 78  
  79 + /**
  80 + * Sets new tagset for this environment.
  81 + *
  82 + * @param tagset
  83 + */
47 84 void setTagset(const Tagset& tagset);
  85 +
  86 + /**
  87 + * Gets currently used tagset.
  88 + *
  89 + * @return
  90 + */
48 91 const Tagset& getTagset() const;
49 92  
  93 + /**
  94 + * Sets binary dictionary file used by this environment.
  95 + *
  96 + * @param filename - filename of the dictionary
  97 + */
50 98 void setFSAFile(const std::string& filename);
51 99  
  100 + /**
  101 + * Sets segmentation rules option.
  102 + *
  103 + * @param option
  104 + * @param value
  105 + */
52 106 void setSegrulesOption(const std::string& option, const std::string& value);
53 107  
  108 + /**
  109 + * Gets segmentation rules automaton.
  110 + *
  111 + * @return
  112 + */
54 113 const SegrulesFSA& getCurrentSegrulesFSA() const;
55 114  
  115 + /**
  116 + * Gets dictionary automaton.
  117 + *
  118 + * @return
  119 + */
56 120 const FSAType& getFSA() const;
57 121  
  122 + /**
  123 + * Returns decoder that converts interpretations to external format.
  124 + * @return
  125 + */
58 126 const InterpretedChunksDecoder& getInterpretedChunksDecoder() const;
59 127  
  128 + /**
  129 + * Gets processor type (info if this is analyzer or generator environment)
  130 + * @return
  131 + */
60 132 MorfeuszProcessorType getProcessorType() const;
61 133  
  134 + /**
  135 + * Return current case pattern helper
  136 + *
  137 + * @return
  138 + */
62 139 const CasePatternHelper& getCasePatternHelper() const;
63 140  
  141 + /**
  142 + * Return current qualifiers helper.
  143 + * @return
  144 + */
64 145 const Qualifiers& getQualifiersHelper() const;
65 146  
  147 + /**
  148 + * Returns true iff given codepoint denotes a separator char for ign handling.
  149 + * @param codepoint
  150 + * @return
  151 + */
66 152 bool isSeparator(uint32_t codepoint) const;
67 153  
68 154 virtual ~Environment();
... ...
morfeusz/InflexionGraph.hpp
  View file @a65ee05
... ... @@ -13,6 +13,10 @@
13 13 #include <utility>
14 14 #include "InterpretedChunk.hpp"
15 15  
  16 +/**
  17 + * This class build inflection graph (indexes the nodes, takes into account segments marked as "weak").
  18 + * Takes care to make the number of nodes as little as possible.
  19 + */
16 20 class InflexionGraph {
17 21 public:
18 22  
... ... @@ -24,30 +28,60 @@ public:
24 28 InterpretedChunk chunk;
25 29 unsigned int nextNode;
26 30 };
27   -
  31 +
  32 + /**
  33 + * Adds new path to the graph.
  34 + *
  35 + * @param path
  36 + * @param weak
  37 + */
28 38 void addPath(const std::vector<InterpretedChunk>& path, bool weak);
29 39  
30 40 // void getResults(const Tagset& tagset, const CharsetConverter& charsetConverter, std::vector<MorphInterpretation>& results);
31 41  
  42 + /**
  43 + * Return current graph.
  44 + *
  45 + * @return
  46 + */
32 47 const std::vector< std::vector<InflexionGraph::Edge> >& getTheGraph();
33 48  
  49 + /**
  50 + * True iff the graph is empty.
  51 + *
  52 + * @return
  53 + */
34 54 bool empty() const;
35 55  
  56 + /**
  57 + * Clears the graph.
  58 + */
36 59 void clear();
37 60  
38   -
39   - // virtual ~FlexionGraph();
40 61 private:
41 62  
42 63 typedef std::pair<const char*, int> PathElement;
43 64 typedef std::set<PathElement> Path;
44 65  
  66 + /**
  67 + * Adds an edge that starts a chunk.
  68 + *
  69 + * @param e
  70 + */
45 71 void addStartEdge(const Edge& e);
46   -
  72 +
  73 + /**
  74 + * Adds non-starting edge.
  75 + * @param startNode
  76 + * @param e
  77 + */
47 78 void addMiddleEdge(unsigned int startNode, const Edge& e);
48 79  
  80 + /**
  81 + * Minimizes the graph so it contains as little number of nodes as possible.
  82 + */
49 83 void minimizeGraph();
50   -
  84 +
51 85 bool canMergeNodes(unsigned int node1, unsigned int node2);
52 86  
53 87 void doMergeNodes(unsigned int node1, unsigned int node2);
... ...
morfeusz/InterpretedChunk.hpp
  View file @a65ee05
... ... @@ -11,16 +11,59 @@
11 11 #include <vector>
12 12 #include "InterpsGroup.hpp"
13 13  
  14 +/**
  15 + * Denotes a part of text that has some not-yet-deserialized interpretations attached to it.
  16 + */
14 17 struct InterpretedChunk {
  18 +
  19 + /**
  20 + * The type of segment for this chunk.
  21 + */
15 22 unsigned char segmentType;
  23 +
  24 + /**
  25 + * Pointer to start of this chunks text
  26 + */
16 27 const char* textStartPtr;
  28 +
  29 + /**
  30 + * Pointer to end of this chunks text (exclusive)
  31 + */
17 32 const char* textEndPtr;
  33 +
  34 + /**
  35 + * Pointer to the start of this chunks binary data.
  36 + */
18 37 const unsigned char* interpsGroupPtr;
  38 +
  39 + /**
  40 + * Pointer to the end of this chunks binary data (exclusive)
  41 + */
19 42 const unsigned char* interpsEndPtr;
  43 +
  44 + /**
  45 + * true iff this chunk shifts orth to the one right to it (it is "A" in "A> B")
  46 + */
20 47 bool shiftOrth;
  48 +
  49 + /**
  50 + * true iff this chunk has attached data from its prefix chunk (when it is "B" segment in "A> B" segmentation rule)
  51 + */
21 52 bool orthWasShifted;
  53 +
  54 + /**
  55 + * Number of codepoints this chunks consists of.
  56 + */
22 57 int codepointsNum;
  58 +
  59 + /**
  60 + * Chunks that are in the prefix segments (those with ">" in segmentation rules, ie. "dig>* dig")
  61 + */
23 62 std::vector<InterpretedChunk> prefixChunks;
  63 +
  64 + /**
  65 + * Homonym id specified by the user.
  66 + */
24 67 std::string requiredHomonymId;
25 68 };
26 69  
... ...
morfeusz/InterpsGroup.hpp
  View file @a65ee05
... ... @@ -10,6 +10,10 @@
10 10  
11 11 #include <stdint.h>
12 12  
  13 +/**
  14 + * A structure representing one segment
  15 + * with pointer to its interpretations, case patterns etc.
  16 + */
13 17 struct InterpsGroup {
14 18 unsigned char type;
15 19 uint16_t size;
... ...
morfeusz/Morfeusz.cpp
  View file @a65ee05
... ... @@ -11,15 +11,13 @@
11 11 #include "utils.hpp"
12 12 #include "data/default_fsa.hpp"
13 13 #include "Morfeusz.hpp"
14   -#include "decoder/InterpretedChunksDecoder.hpp"
  14 +#include "deserialization/morphInterps/InterpretedChunksDecoder.hpp"
15 15 #include "charset/CharsetConverter.hpp"
16 16 #include "charset/charset_utils.hpp"
17   -#include "charset/CaseConverter.hpp"
  17 +#include "case/CaseConverter.hpp"
18 18 #include "segrules/segrules.hpp"
19 19 #include "const.hpp"
20   -#include "deserializationUtils.hpp"
21 20 #include "charset/utf8.h"
22   -#include "compressionByteUtils.hpp"
23 21  
24 22 // TODO - konstruktor kopiujący działający Tak-Jak-Trzeba
25 23  
... ...
morfeusz/Morfeusz.hpp
  View file @a65ee05
... ... @@ -13,12 +13,11 @@
13 13 #include <vector>
14 14 #include <map>
15 15 #include <set>
16   -#include "EncodedInterpretation.hpp"
17 16 #include "fsa/fsa.hpp"
18 17 #include "MorphInterpretation.hpp"
19 18 #include "InterpsGroup.hpp"
  19 +#include "case/CaseConverter.hpp"
20 20 #include "charset/CharsetConverter.hpp"
21   -#include "charset/CaseConverter.hpp"
22 21 #include "charset/TextReader.hpp"
23 22 #include "InterpretedChunk.hpp"
24 23 #include "InflexionGraph.hpp"
... ... @@ -30,8 +29,8 @@
30 29 #include "segrules/segrules.hpp"
31 30 #include "segrules/SegrulesFSA.hpp"
32 31  
33   -#include "deserializer/InterpsGroupsReader.hpp"
34   -#include "deserializer/MorphDeserializer.hpp"
  32 +#include "deserialization/InterpsGroupsReader.hpp"
  33 +#include "deserialization/MorphDeserializer.hpp"
35 34  
36 35 class Morfeusz;
37 36 class ResultsIterator;
... ...
morfeusz/MorfeuszOptions.hpp
  View file @a65ee05
... ... @@ -10,6 +10,9 @@
10 10  
11 11 #include "const.hpp"
12 12  
  13 +/**
  14 + * Represents options for Morfeusz analyzer/generator.
  15 + */
13 16 struct MorfeuszOptions {
14 17 bool caseSensitive;
15 18 MorfeuszCharset encoding;
... ...
morfeusz/MorfeuszConfig.hpp.in renamed to morfeusz/MorfeuszVersion.hpp.in
  View file @a65ee05
... ... @@ -5,6 +5,4 @@
5 5 * Created on November 29, 2013, 10:03 PM
6 6 */
7 7  
8   -#define Morfeusz_VERSION_MAJOR @Morfeusz_VERSION_MAJOR@
9   -#define Morfeusz_VERSION_MINOR @Morfeusz_VERSION_MINOR@
10   -
  8 +#define MORFEUSZ_VERSION "@Morfeusz_VERSION@"
... ...
morfeusz/MorphDeserializer.cpp deleted
View file @a1d73c9
1   -/*
2   - * File: MorphDeserializer.cpp
3   - * Author: mlenart
4   - *
5   - * Created on 12 listopad 2013, 15:31
6   - */
7   -
8   -#include <map>
9   -#include <algorithm>
10   -#include "MorphDeserializer.hpp"
11   -#include "EncodedInterpretation.hpp"
12   -#include "InterpsGroup.hpp"
13   -#include "deserializationUtils.hpp"
14   -
15   -MorphDeserializer::MorphDeserializer() {
16   -}
17   -
18   -MorphDeserializer::~MorphDeserializer() {
19   -}
20   -
21   -long MorphDeserializer::deserialize(const unsigned char* ptr, vector<InterpsGroup>& interps) const {
22   - const unsigned char* currPtr = ptr;
23   - uint8_t interpTypesNum = readInt8(currPtr);
24   - interps.clear();
25   - interps.reserve(interpTypesNum);
26   - for (unsigned int i = 0; i < interpTypesNum; i++) {
27   - InterpsGroup ig;
28   - ig.type = readInt8(currPtr);
29   - ig.size = readInt16(currPtr);
30   - ig.ptr = currPtr;
31   - currPtr += ig.size;
32   - interps.push_back(ig);
33   - }
34   - return currPtr - ptr;
35   -}
morfeusz/MorphInterpretation.cpp
  View file @a65ee05
... ... @@ -8,7 +8,6 @@
8 8 #include <string>
9 9 #include <sstream>
10 10 #include "MorphInterpretation.hpp"
11   -#include "EncodedInterpretation.hpp"
12 11 #include "const.hpp"
13 12  
14 13 using namespace std;
... ... @@ -32,11 +31,13 @@ tagnum(tagnum),
32 31 namenum(namenum),
33 32 tag(env.getTagset().getTag(tagnum, env.getCharsetConverter())),
34 33 name(env.getTagset().getName(namenum, env.getCharsetConverter())),
35   -qualifiers(env.getQualifiersHelper().getQualifiers(qualifiersNum)) {
  34 +qualifiers(&env.getQualifiersHelper().getQualifiers(qualifiersNum)) {
36 35  
37 36  
38 37 }
39 38  
  39 +static const vector<std::string> emptyQualifiers;
  40 +
40 41 MorphInterpretation::MorphInterpretation()
41 42 : startNode(),
42 43 endNode(),
... ... @@ -47,7 +48,7 @@ tagnum(),
47 48 namenum(),
48 49 tag(),
49 50 name(),
50   -qualifiers(){
  51 +qualifiers(&emptyQualifiers){
51 52  
52 53 }
53 54  
... ... @@ -65,7 +66,7 @@ namenum(0),
65 66 // qualifiersNum(0),
66 67 tag(env.getTagset().getTag(0, env.getCharsetConverter())),
67 68 name(env.getTagset().getName(0, env.getCharsetConverter())),
68   -qualifiers() {
  69 +qualifiers(&emptyQualifiers) {
69 70  
70 71 }
71 72  
... ... @@ -126,7 +127,7 @@ const std::string&amp; MorphInterpretation::getName() const {
126 127 }
127 128  
128 129 const vector<string>& MorphInterpretation::getQualifiers() const {
129   - return this->qualifiers;
  130 + return *this->qualifiers;
130 131 }
131 132  
132 133 static inline string getQualifiersStr(const MorphInterpretation& mi) {
... ... @@ -157,7 +158,7 @@ std::string MorphInterpretation::toString(bool includeNodeNumbers) const {
157 158 if (!name.empty()) {
158 159 res << "," << name;
159 160 }
160   - if (!qualifiers.empty()) {
  161 + if (!qualifiers->empty()) {
161 162 res << "," << getQualifiersStr(*this);
162 163 }
163 164 return res.str();
... ...
morfeusz/MorphInterpretation.hpp
  View file @a65ee05
... ... @@ -13,10 +13,12 @@
13 13 class Environment;
14 14  
15 15 #include "Tagset.hpp"
16   -#include "EncodedInterpretation.hpp"
17 16 #include "charset/CharsetConverter.hpp"
18 17 #include "Environment.hpp"
19 18  
  19 +/**
  20 + * Morphological interpretation as seen by the user in the analysis/generation results.
  21 + */
20 22 class MorphInterpretation {
21 23 public:
22 24 MorphInterpretation(
... ... @@ -59,7 +61,7 @@ private:
59 61 int namenum;
60 62 std::string tag;
61 63 std::string name;
62   - std::vector<std::string> qualifiers;
  64 + const std::vector<std::string>* qualifiers;
63 65 };
64 66  
65 67 #endif /* MORPHINTERPRETATION_HPP */
... ...
morfeusz/Qualifiers.cpp
  View file @a65ee05
... ... @@ -7,7 +7,7 @@
7 7  
8 8 #include <iostream>
9 9 #include "Qualifiers.hpp"
10   -#include "deserializationUtils.hpp"
  10 +#include "deserialization/deserializationUtils.hpp"
11 11 #include "fsa/const.hpp"
12 12  
13 13 using namespace std;
... ... @@ -32,13 +32,9 @@ qualifiers() {
32 32 }
33 33 }
34 34  
35   -vector<string> Qualifiers::getQualifiers(int n) const {
  35 +const vector<string>& Qualifiers::getQualifiers(int n) const {
36 36 return this->qualifiers.at(n);
37 37 }
38 38  
39   -unsigned int Qualifiers::getQualifiersNum() const {
40   - return (unsigned int) this->qualifiers.size();
41   -}
42   -
43 39 Qualifiers::~Qualifiers() {
44 40 }
... ...
morfeusz/Qualifiers.hpp
  View file @a65ee05
... ... @@ -12,11 +12,21 @@
12 12 #include <string>
13 13 #include <stdint.h>
14 14  
  15 +/**
  16 + * Helper class used for decoding qualifiers set number into a vector of strings.
  17 + *
  18 + * @param ptr
  19 + */
15 20 class Qualifiers {
16 21 public:
17 22 explicit Qualifiers(const unsigned char* ptr);
18   - std::vector<std::string> getQualifiers(int n) const;
19   - unsigned int getQualifiersNum() const;
  23 +
  24 + /**
  25 + * Returns vector of qualifiers represented as strings.
  26 + * @param n - the index in qualifiers tab.
  27 + * @return - vector of qualifiers represented as strings.
  28 + */
  29 + const std::vector<std::string>& getQualifiers(int n) const;
20 30 virtual ~Qualifiers();
21 31 private:
22 32 std::vector< std::vector<std::string> > qualifiers;
... ...
morfeusz/Tagset.cpp
  View file @a65ee05
... ... @@ -3,8 +3,7 @@
3 3 #include "Tagset.hpp"
4 4 #include "fsa/const.hpp"
5 5 #include "utils.hpp"
6   -#include "endianness.hpp"
7   -#include "deserializationUtils.hpp"
  6 +#include "deserialization/deserializationUtils.hpp"
8 7  
9 8 using namespace std;
10 9  
... ...
morfeusz/Tagset.hpp
  View file @a65ee05
... ... @@ -12,11 +12,34 @@
12 12 #include <vector>
13 13 #include "charset/CharsetConverter.hpp"
14 14  
  15 +/**
  16 + * Represents a tagset
  17 + */
15 18 class Tagset {
16 19 public:
  20 + /**
  21 + * Constructs a tagset from binary data.
  22 + *
  23 + * @param fsaData - pointer to the beginning of automaton data.
  24 + */
17 25 explicit Tagset(const unsigned char* fsaData);
18   -// Tagset(const Tagset& tagset);
  26 +
  27 + /**
  28 + * Returns tag (denoted by its index) as a string.
  29 + *
  30 + * @param tagNum - tag index in the tagset.
  31 + * @param charsetConverter - the charset converter used to convert from tagset internal encoding (UTF-8) into target encoding.
  32 + * @return - the tag encoded as string.
  33 + */
19 34 const std::string getTag(const int tagNum, const CharsetConverter& charsetConverter) const;
  35 +
  36 + /**
  37 + * Returns named entity type (denoted by its index) as a string.
  38 + *
  39 + * @param nameNum - name index in the tagset.
  40 + * @param charsetConverter - the charset converter used to convert from tagset internal encoding (UTF-8) into target encoding.
  41 + * @return - the named entity type encoded as string.
  42 + */
20 43 const std::string getName(const int nameNum, const CharsetConverter& charsetConverter) const;
21 44 private:
22 45 std::vector<std::string> tags;
... ...
morfeusz/charset/CaseConverter.cpp renamed to morfeusz/case/CaseConverter.cpp
  View file @a65ee05
morfeusz/charset/CaseConverter.hpp renamed to morfeusz/case/CaseConverter.hpp
  View file @a65ee05
morfeusz/CasePatternHelper.cpp renamed to morfeusz/case/CasePatternHelper.cpp
  View file @a65ee05
morfeusz/CasePatternHelper.hpp renamed to morfeusz/case/CasePatternHelper.hpp
  View file @a65ee05
... ... @@ -10,12 +10,15 @@
10 10  
11 11 #include <vector>
12 12 #include "InterpsGroup.hpp"
13   -#include "CasePatternHelper.hpp"
14   -#include "compressionByteUtils.hpp"
  13 +#include "deserialization/morphInterps/compressionByteUtils.hpp"
15 14 #include "Environment.hpp"
16 15  
17 16 class Environment;
18 17  
  18 +/**
  19 + * Utility class used to for case-sensitive interpretations filtering
  20 + * (ie. to filter out "berlin" and keep "Berlin")
  21 + */
19 22 class CasePatternHelper {
20 23 public:
21 24  
... ... @@ -23,10 +26,23 @@ public:
23 26  
24 27 }
25 28  
  29 + /**
  30 + * Set if this case pattern helper cares about case-sensitivity
  31 + *
  32 + * @param caseSensitive
  33 + */
26 34 void setCaseSensitive(bool caseSensitive) {
27 35 this->caseSensitive = caseSensitive;
28 36 }
29   -
  37 +
  38 + /**
  39 + * Check if given word matches given case pattern
  40 + *
  41 + * @param lowercaseCodepoints - codepoints of checked word converter to lowercase
  42 + * @param originalCodepoints - codepoints of checked word
  43 + * @param casePattern - vector representing case pattern ( ie. [False, True] for "mBank")
  44 + * @return - true iff word denoted by given codepoints matches given case pattern
  45 + */
30 46 bool checkCasePattern(
31 47 const std::vector<uint32_t>& lowercaseCodepoints,
32 48 const std::vector<uint32_t>& originalCodepoints,
... ... @@ -41,12 +57,28 @@ public:
41 57 return true;
42 58 }
43 59  
  60 + /**
  61 + * Check if given word has a chance of matching any of case patterns in given interps group.
  62 + *
  63 + * @param env - environment
  64 + * @param orthStart - pointer to start of word
  65 + * @param orthEnd - pointer to end of word
  66 + * @param ig - interps group
  67 + * @return - true iff word encoded from orthStart to orthEnd
  68 + * matches at least one of the interp group's morph interpretation's case pattern.
  69 + */
44 70 bool checkInterpsGroupOrthCasePatterns(
45 71 const Environment& env,
46 72 const char* orthStart,
47 73 const char* orthEnd,
48 74 const InterpsGroup& ig) const;
49 75  
  76 + /**
  77 + * Deserializes case pattern encoded at given pointer.
  78 + *
  79 + * @param ptr
  80 + * @return - case pattern
  81 + */
50 82 static std::vector<bool> deserializeOneCasePattern(const unsigned char*& ptr);
51 83 private:
52 84 bool caseSensitive;
... ...
morfeusz/charset/caseconv.cpp renamed to morfeusz/case/caseconv.cpp
  View file @a65ee05
morfeusz/charset/caseconv.hpp renamed to morfeusz/case/caseconv.hpp
  View file @a65ee05
... ... @@ -8,6 +8,10 @@
8 8 #ifndef CASECONV_HPP
9 9 #define CASECONV_HPP
10 10  
  11 +/*
  12 + * Case conversion tables
  13 + */
  14 +
11 15 extern const unsigned int TO_LOWERCASE_TABLE_SIZE;
12 16 extern const unsigned int EXT_TO_LOWERCASE_TABLE_SIZE;
13 17 extern const uint32_t TO_LOWERCASE_TABLE[];
... ...
morfeusz/charset/CharsetConverter.cpp
  View file @a65ee05
... ... @@ -4,7 +4,7 @@
4 4 #include <algorithm>
5 5 #include <inttypes.h>
6 6 #include <iostream>
7   -#include "../endianness.hpp"
  7 +#include "deserialization/endianness.hpp"
8 8 #include "utf8.h"
9 9 #include "CharsetConverter.hpp"
10 10 #include "conversion_tables.hpp"
... ...
morfeusz/charset/TextReader.hpp
  View file @a65ee05
... ... @@ -8,7 +8,7 @@
8 8 #ifndef TEXTREADER_HPP
9 9 #define TEXTREADER_HPP
10 10  
11   -#include "../Environment.hpp"
  11 +#include "Environment.hpp"
12 12  
13 13 class TextReader {
14 14 public:
... ...
morfeusz/outputUtils.hpp renamed to morfeusz/cli/outputUtils.hpp
  View file @a65ee05
morfeusz/deserializer/InterpsGroupsReader.cpp renamed to morfeusz/deserialization/InterpsGroupsReader.cpp
  View file @a65ee05
... ... @@ -6,7 +6,7 @@
6 6 */
7 7  
8 8 #include "InterpsGroupsReader.hpp"
9   -#include "../deserializationUtils.hpp"
  9 +#include "deserialization/deserializationUtils.hpp"
10 10  
11 11 InterpsGroupsReader::InterpsGroupsReader()
12 12 : currPtr(NULL), endPtr(NULL) {
... ...
morfeusz/deserializer/InterpsGroupsReader.hpp renamed to morfeusz/deserialization/InterpsGroupsReader.hpp
  View file @a65ee05
morfeusz/deserializer/MorphDeserializer.cpp renamed to morfeusz/deserialization/MorphDeserializer.cpp
  View file @a65ee05
... ... @@ -6,7 +6,7 @@
6 6 */
7 7  
8 8 #include "MorphDeserializer.hpp"
9   -#include "../deserializationUtils.hpp"
  9 +#include "deserialization/deserializationUtils.hpp"
10 10  
11 11 MorphDeserializer::MorphDeserializer() {
12 12 }
... ...
morfeusz/deserializer/MorphDeserializer.hpp renamed to morfeusz/deserialization/MorphDeserializer.hpp
  View file @a65ee05
morfeusz/deserializationUtils.hpp renamed to morfeusz/deserialization/deserializationUtils.hpp
  View file @a65ee05
morfeusz/endianness.hpp renamed to morfeusz/deserialization/endianness.hpp
  View file @a65ee05
morfeusz/EncodedInterpretation.hpp renamed to morfeusz/deserialization/morphInterps/EncodedInterpretation.hpp
  View file @a65ee05
morfeusz/decoder/InterpretedChunksDecoder.cpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder.cpp
  View file @a65ee05
morfeusz/decoder/InterpretedChunksDecoder.hpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder.hpp
  View file @a65ee05
... ... @@ -16,11 +16,10 @@
16 16 #include "EncodedInterpretation.hpp"
17 17 #include "InterpretedChunk.hpp"
18 18 #include "EncodedInterpretation.hpp"
19   -#include "charset/CaseConverter.hpp"
  19 +#include "case/CaseConverter.hpp"
20 20 #include "Environment.hpp"
21 21 #include "MorphInterpretation.hpp"
22   -#include "CasePatternHelper.hpp"
23   -#include "deserializationUtils.hpp"
  22 +#include "case/CasePatternHelper.hpp"
24 23 #include "compressionByteUtils.hpp"
25 24 #include "const.hpp"
26 25  
... ...
morfeusz/decoder/InterpretedChunksDecoder4Analyzer.cpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.cpp
  View file @a65ee05
morfeusz/decoder/InterpretedChunksDecoder4Analyzer.hpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.hpp
  View file @a65ee05
morfeusz/decoder/InterpretedChunksDecoder4Generator.cpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Generator.cpp
  View file @a65ee05
morfeusz/decoder/InterpretedChunksDecoder4Generator.hpp renamed to morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Generator.hpp
  View file @a65ee05
morfeusz/compressionByteUtils.hpp renamed to morfeusz/deserialization/morphInterps/compressionByteUtils.hpp
  View file @a65ee05
morfeusz/fsa/cfsa1_impl.hpp
  View file @a65ee05
... ... @@ -12,7 +12,7 @@
12 12 #include <climits>
13 13  
14 14 #include "fsa.hpp"
15   -#include "../deserializationUtils.hpp"
  15 +#include "../deserialization/deserializationUtils.hpp"
16 16  
17 17 static const unsigned char CFSA1_ACCEPTING_FLAG = 128;
18 18 //static const unsigned char CFSA1_ARRAY_FLAG = 64;
... ...
morfeusz/fsa/cfsa2_impl.hpp
  View file @a65ee05
... ... @@ -13,7 +13,7 @@
13 13 #include <iostream>
14 14 #include "fsa.hpp"
15 15 #include "../utils.hpp"
16   -#include "../endianness.hpp"
  16 +#include "../deserialization/endianness.hpp"
17 17  
18 18 static const unsigned char HAS_REMAINING_FLAG = 128;
19 19 static const unsigned char ACCEPTING_FLAG = 64;
... ...
morfeusz/fsa/fsa_impl.hpp
  View file @a65ee05
... ... @@ -17,7 +17,7 @@
17 17 #include <sstream>
18 18 #include "const.hpp"
19 19 #include "../utils.hpp"
20   -#include "../endianness.hpp"
  20 +#include "../deserialization/endianness.hpp"
21 21  
22 22 //using namespace std;
23 23 //static const unsigned int FSA_OFFSET = 6;
... ...
morfeusz/java/dupa deleted
View file @a1d73c9
1   -
2   -# SWIG
3   -#set(CMAKE_SWIG_OUTDIR swig)
4   -#FIND_PACKAGE(SWIG REQUIRED)
5   -FIND_PACKAGE(JNI REQUIRED)
6   -#INCLUDE(${SWIG_USE_FILE})
7   -include(UseJava)
8   -
9   -# SWIG Java
10   -include_directories (${JAVA_INCLUDE_PATH})
11   -include_directories (..)
12   -
13   -set (SWIG_JAVA_OUTFILE swigJAVA.cpp)
14   -# set (JAVA_WRAPPER_FILE ${CMAKE_SHARED_LIBRARY_PREFIX}morfeusz${CMAKE_SHARED_LIBRARY_SUFFIX})
15   -add_custom_command (
16   - OUTPUT ${SWIG_JAVA_OUTFILE}
17   - COMMAND swig -java -c++ -package pl.waw.ipipan.morfeusz -o ${SWIG_JAVA_OUTFILE} -outdir ${CMAKE_SOURCE_DIR}/jmorfeusz/src/main/java/pl/waw/ipipan/morfeusz ${CMAKE_SOURCE_DIR}/morfeusz/morfeusz.i
18   - DEPENDS libmorfeusz
19   -)
20   -#set (CMAKE_SHARED_LINKER_FLAGS "-s -Os -static-libstdc++ -static-libgcc")
21   -add_library (jmorfeusz SHARED ${SWIG_JAVA_OUTFILE})
22   -target_link_libraries (jmorfeusz ${JAVA_LIBRARIES} libmorfeusz)
23   -add_dependencies (jmorfeusz ${SWIG_JAVA_OUTFILE})
24   -
25   -#set (CMAKE_SWIG_FLAGS -package pl.waw.ipipan.morfeusz)
26   -#set (CMAKE_SWIG_OUTDIR ${CMAKE_SOURCE_DIR}/jmorfeusz/src/main/java/pl/waw/ipipan/morfeusz)
27   -
28   -#set_source_files_properties (../morfeusz.i PROPERTIES CPLUSPLUS ON)
29   -#SWIG_ADD_MODULE(jmorfeusz java ../morfeusz.i)
30   -#SWIG_LINK_LIBRARIES(jmorfeusz ${JAVA_LIBRARIES})
31   -#SWIG_LINK_LIBRARIES(jmorfeusz libmorfeusz)
32   -
33   -#if (${CMAKE_SYSTEM_NAME} MATCHES "Windows")
34   -# set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
35   -# set (CMAKE_SHARED_LINKER_FLAGS "-s -Os -static-libstdc++ -static-libgcc")
36   -#endif ()
morfeusz/morfeusz_analyzer.cpp
  View file @a65ee05
... ... @@ -12,14 +12,16 @@
12 12 #include "fsa/fsa.hpp"
13 13 #include "Tagset.hpp"
14 14 #include "Morfeusz.hpp"
  15 +#include "MorfeuszVersion.hpp"
15 16 #include "const.hpp"
16 17  
17 18 #include "cli/cli.hpp"
18   -#include "outputUtils.hpp"
  19 +#include "cli/outputUtils.hpp"
19 20  
20 21 using namespace std;
21 22  
22 23 int main(int argc, const char** argv) {
  24 + cerr << "Morfeusz analyzer, version: " << MORFEUSZ_VERSION << endl;
23 25 ez::ezOptionParser& opt = *getOptions(argc, argv, ANALYZER);
24 26 Morfeusz morfeusz;
25 27 initializeMorfeusz(opt, morfeusz);
... ...
morfeusz/morfeusz_generator.cpp
  View file @a65ee05
... ... @@ -11,13 +11,15 @@
11 11 #include "fsa/fsa.hpp"
12 12 #include "Tagset.hpp"
13 13 #include "Morfeusz.hpp"
  14 +#include "MorfeuszVersion.hpp"
14 15 #include "const.hpp"
15 16 #include "cli/cli.hpp"
16   -#include "outputUtils.hpp"
  17 +#include "cli/outputUtils.hpp"
17 18  
18 19 using namespace std;
19 20  
20 21 int main(int argc, const char** argv) {
  22 + cerr << "Morfeusz generator, version: " << MORFEUSZ_VERSION << endl;
21 23 ez::ezOptionParser& opt = *getOptions(argc, argv, GENERATOR);
22 24 Morfeusz morfeusz;
23 25 initializeMorfeusz(opt, morfeusz);
... ...
morfeusz/segrules/SegrulesFSA.hpp
  View file @a65ee05
... ... @@ -10,7 +10,7 @@
10 10  
11 11 #include <set>
12 12 #include <iostream>
13   -#include "../deserializationUtils.hpp"
  13 +#include "../deserialization/deserializationUtils.hpp"
14 14  
15 15 struct SegrulesState {
16 16 uint16_t offset;
... ...
morfeusz/segrules/segrules.cpp
  View file @a65ee05
1 1  
2 2 #include "segrules.hpp"
3   -#include "../fsa/fsa.hpp"
4   -#include "../fsa/const.hpp"
5   -#include "../deserializationUtils.hpp"
  3 +#include "fsa/fsa.hpp"
  4 +#include "fsa/const.hpp"
  5 +#include "deserialization/deserializationUtils.hpp"
6 6  
7 7 using namespace std;
8 8  
... ...
morfeusz/consoleUtils.hpp renamed to morfeusz/test/consoleUtils.hpp
  View file @a65ee05
morfeusz/test_recognize_dict.cpp renamed to morfeusz/test/test_recognize_dict.cpp
  View file @a65ee05
... ... @@ -8,7 +8,6 @@
8 8 //#include <cstdlib>
9 9 #include <sstream>
10 10 #include <iostream>
11   -#include "EncodedInterpretation.hpp"
12 11 #include "utils.hpp"
13 12 #include "Morfeusz.hpp"
14 13 #include "MorphInterpretation.hpp"
... ...
morfeusz/test_result_equals.cpp renamed to morfeusz/test/test_result_equals.cpp
  View file @a65ee05
morfeusz/test_synth_dict.cpp deleted
View file @a1d73c9
1   -/*
2   - * File: test_synth_dict.cpp
3   - * Author: mlenart
4   - *
5   - * Created on 21 styczeń 2014, 12:00
6   - */
7   -
8   -#include <cstdlib>
9   -
10   -using namespace std;
11   -
12   -/*
13   - *
14   - */
15   -int main(int argc, char** argv) {
16   -
17   - return 0;
18   -}
19   -
morfeusz/wrappers/CMakeLists.txt 0 → 100644
  View file @a65ee05
  1 +add_subdirectory (java)
  2 +add_subdirectory (python)
  3 +add_subdirectory (perl)
0 4 \ No newline at end of file
... ...
morfeusz/java/CMakeLists.txt renamed to morfeusz/wrappers/java/CMakeLists.txt
  View file @a65ee05
... ... @@ -6,14 +6,14 @@ find_package(Java REQUIRED)
6 6  
7 7 include_directories (${JAVA_INCLUDE_PATH})
8 8 include_directories (${JAVA_INCLUDE_PATH2})
9   -include_directories (..)
  9 +include_directories (${CMAKE_SOURCE_DIR}/morfeusz)
10 10  
11 11 set (SWIG_JAVA_OUTFILE "${CMAKE_CURRENT_BINARY_DIR}/swigJAVA.cpp")
12 12 file (COPY pl DESTINATION .)
13 13 set (JAVA_SRC_DIR "${CMAKE_CURRENT_BINARY_DIR}/pl/waw/ipipan/morfeusz")
14 14 add_custom_command (
15 15 OUTPUT ${SWIG_JAVA_OUTFILE}
16   - COMMAND swig -java -c++ -package pl.waw.ipipan.morfeusz -o ${SWIG_JAVA_OUTFILE} -outdir ${JAVA_SRC_DIR} ${CMAKE_SOURCE_DIR}/morfeusz/morfeusz.i
  16 + COMMAND swig -java -c++ -package pl.waw.ipipan.morfeusz -o ${SWIG_JAVA_OUTFILE} -outdir ${JAVA_SRC_DIR} ${CMAKE_SOURCE_DIR}/morfeusz/wrappers/morfeusz.i
17 17 DEPENDS libmorfeusz
18 18 )
19 19 add_custom_target(generate_java_wrapper ALL
... ...
morfeusz/java/README renamed to morfeusz/wrappers/java/README
  View file @a65ee05
morfeusz/java/pl/waw/ipipan/morfeusz/app/App.java renamed to morfeusz/wrappers/java/pl/waw/ipipan/morfeusz/app/App.java
  View file @a65ee05
morfeusz/java/pl/waw/ipipan/morfeusz/app/MorfeuszUtils.java renamed to morfeusz/wrappers/java/pl/waw/ipipan/morfeusz/app/MorfeuszUtils.java
  View file @a65ee05
morfeusz/morfeusz.i renamed to morfeusz/wrappers/morfeusz.i
  View file @a65ee05
... ... @@ -91,10 +91,10 @@ import java.io.IOException;
91 91  
92 92 %ignore Tagset::Tagset(const unsigned char* fsaData);
93 93  
94   -%include "Morfeusz.hpp"
95   -%include "MorphInterpretation.hpp"
96   -%include "const.hpp"
97   -%include "exceptions.hpp"
  94 +%include "../Morfeusz.hpp"
  95 +%include "../MorphInterpretation.hpp"
  96 +%include "../const.hpp"
  97 +%include "../exceptions.hpp"
98 98  
99 99 // instantiate vector of interpretations
100 100 namespace std {
... ...
morfeusz/perl/CMakeLists.txt renamed to morfeusz/wrappers/perl/CMakeLists.txt
  View file @a65ee05
... ... @@ -6,7 +6,7 @@ if (NOT CMAKE_CROSSCOMPILING)
6 6  
7 7 find_package (PerlLibs REQUIRED)
8 8 include_directories (${PERL_INCLUDE_PATH})
9   - include_directories (..)
  9 + include_directories (../..)
10 10  
11 11 set (CMAKE_SWIG_FLAGS "")
12 12  
... ...
morfeusz/python/CMakeLists.txt renamed to morfeusz/wrappers/python/CMakeLists.txt
  View file @a65ee05
... ... @@ -9,14 +9,14 @@ set (PYMORFEUSZ_VERSION &quot;0.1.0&quot;)
9 9  
10 10 # SWIG Java
11 11 INCLUDE_DIRECTORIES (${PYTHON_INCLUDE_PATH})
12   -INCLUDE_DIRECTORIES (..)
  12 +INCLUDE_DIRECTORIES (../..)
13 13  
14 14  
15 15 set (SWIG_PYTHON_OUTFILE_CXX "${CMAKE_CURRENT_BINARY_DIR}/swigPYTHON.cpp")
16 16 set (SWIG_PYTHON_OUTFILE_PY "${CMAKE_CURRENT_BINARY_DIR}/morfeusz2.py")
17 17 add_custom_command (
18 18 OUTPUT "${SWIG_PYTHON_OUTFILE_CXX}" "${SWIG_PYTHON_OUTFILE_PY}"
19   - COMMAND swig -python -c++ -o "${SWIG_PYTHON_OUTFILE_CXX}" "${CMAKE_SOURCE_DIR}/morfeusz/morfeusz.i"
  19 + COMMAND swig -python -c++ -o "${SWIG_PYTHON_OUTFILE_CXX}" "${CMAKE_SOURCE_DIR}/morfeusz/wrappers/morfeusz.i"
20 20 DEPENDS libmorfeusz
21 21 )
22 22 add_custom_target (generate_python_wrapper
... ...
morfeusz/python/setup.py.in renamed to morfeusz/wrappers/python/setup.py.in
  View file @a65ee05
nbproject/configurations.xml
  View file @a65ee05
... ... @@ -5,6 +5,14 @@
5 5 displayName="build"
6 6 projectFiles="true"
7 7 root="build">
  8 + <logicalFolder name="morfeusz" displayName="morfeusz" projectFiles="true">
  9 + <logicalFolder name="wrappers" displayName="wrappers" projectFiles="true">
  10 + <logicalFolder name="java" displayName="java" projectFiles="true">
  11 + <itemPath>build/morfeusz/wrappers/java/swigJAVA.cpp</itemPath>
  12 + </logicalFolder>
  13 + <itemPath>build/morfeusz/wrappers/morfeuszPERL_wrap.cxx</itemPath>
  14 + </logicalFolder>
  15 + </logicalFolder>
8 16 <itemPath>build/default_fsa.cpp</itemPath>
9 17 <itemPath>build/default_synth_fsa.cpp</itemPath>
10 18 </logicalFolder>
... ... @@ -17,22 +25,25 @@
17 25 <itemPath>build1/morfeusz/java/swigJAVA.cpp</itemPath>
18 26 </logicalFolder>
19 27 <df root="morfeusz" name="0">
20   - <df name="charset">
  28 + <df name="case">
21 29 <in>CaseConverter.cpp</in>
  30 + <in>CasePatternHelper.cpp</in>
  31 + <in>caseconv.cpp</in>
  32 + </df>
  33 + <df name="charset">
22 34 <in>CharsetConverter.cpp</in>
23 35 <in>TextReader.cpp</in>
24   - <in>caseconv.cpp</in>
25 36 <in>conversion_tables.cpp</in>
26 37 </df>
27 38 <df name="cli">
28 39 <in>cli.cpp</in>
29 40 </df>
30   - <df name="decoder">
31   - <in>InterpretedChunksDecoder.cpp</in>
32   - <in>InterpretedChunksDecoder4Analyzer.cpp</in>
33   - <in>InterpretedChunksDecoder4Generator.cpp</in>
34   - </df>
35   - <df name="deserializer">
  41 + <df name="deserialization">
  42 + <df name="morphInterps">
  43 + <in>InterpretedChunksDecoder.cpp</in>
  44 + <in>InterpretedChunksDecoder4Analyzer.cpp</in>
  45 + <in>InterpretedChunksDecoder4Generator.cpp</in>
  46 + </df>
36 47 <in>InterpsGroupsReader.cpp</in>
37 48 <in>MorphDeserializer.cpp</in>
38 49 </df>
... ... @@ -46,20 +57,19 @@
46 57 <in>SegrulesFSA.cpp</in>
47 58 <in>segrules.cpp</in>
48 59 </df>
49   - <in>CasePatternHelper.cpp</in>
  60 + <df name="test">
  61 + <in>test_recognize_dict.cpp</in>
  62 + <in>test_result_equals.cpp</in>
  63 + </df>
50 64 <in>Environment.cpp</in>
51 65 <in>InflexionGraph.cpp</in>
52 66 <in>Morfeusz.cpp</in>
53   - <in>MorphDeserializer.cpp</in>
54 67 <in>MorphInterpretation.cpp</in>
55 68 <in>Qualifiers.cpp</in>
56 69 <in>Tagset.cpp</in>
57 70 <in>const.cpp</in>
58   - <in>main.cpp</in>
59 71 <in>morfeusz_analyzer.cpp</in>
60 72 <in>morfeusz_generator.cpp</in>
61   - <in>test_recognize_dict.cpp</in>
62   - <in>test_result_equals.cpp</in>
63 73 </df>
64 74 <logicalFolder name="morfeusz"
65 75 displayName="morfeusz"
... ... @@ -101,10 +111,9 @@
101 111 <rebuildPropChanged>false</rebuildPropChanged>
102 112 </toolsSet>
103 113 <flagsDictionary>
104   - <element flagsID="0" commonFlags="-O2 -std=c++98"/>
105   - <element flagsID="1" commonFlags="-O2 -std=c++98 -fPIC"/>
106   - <element flagsID="2" commonFlags="-std=c++98 -O3"/>
107   - <element flagsID="3" commonFlags="-std=c++98 -O3 -fPIC"/>
  114 + <element flagsID="0" commonFlags="-std=c++98 -O3"/>
  115 + <element flagsID="1" commonFlags="-std=c++98 -O3 -fPIC"/>
  116 + <element flagsID="2" commonFlags="3"/>
108 117 </flagsDictionary>
109 118 <codeAssistance>
110 119 </codeAssistance>
... ... @@ -114,18 +123,49 @@
114 123 <buildCommand>${MAKE} -f Makefile</buildCommand>
115 124 <cleanCommand>${MAKE} -f Makefile clean</cleanCommand>
116 125 <executablePath>build/morfeusz/morfeusz_analyzer</executablePath>
  126 + <ccTool flags="1">
  127 + <incDir>
  128 + <pElem>build</pElem>
  129 + <pElem>morfeusz</pElem>
  130 + </incDir>
  131 + <preprocessorList>
  132 + <Elem>NDEBUG</Elem>
  133 + </preprocessorList>
  134 + </ccTool>
117 135 </makeTool>
118 136 </makefileType>
119 137 <item path="build/default_fsa.cpp" ex="false" tool="1" flavor2="4">
  138 + <ccTool>
  139 + <incDir>
  140 + <pElem>build/morfeusz</pElem>
  141 + </incDir>
  142 + <preprocessorList>
  143 + <Elem>__PIC__=2</Elem>
  144 + <Elem>__pic__=2</Elem>
  145 + <Elem>libmorfeusz_EXPORTS</Elem>
  146 + </preprocessorList>
  147 + <undefinedList>
  148 + <Elem>__GCC_HAVE_DWARF2_CFI_ASM=1</Elem>
  149 + </undefinedList>
  150 + </ccTool>
120 151 </item>
121 152 <item path="build/default_synth_fsa.cpp" ex="false" tool="1" flavor2="4">
  153 + <ccTool>
  154 + <incDir>
  155 + <pElem>build/morfeusz</pElem>
  156 + </incDir>
  157 + <preprocessorList>
  158 + <Elem>__PIC__=2</Elem>
  159 + <Elem>__pic__=2</Elem>
  160 + <Elem>libmorfeusz_EXPORTS</Elem>
  161 + </preprocessorList>
  162 + <undefinedList>
  163 + <Elem>__GCC_HAVE_DWARF2_CFI_ASM=1</Elem>
  164 + </undefinedList>
  165 + </ccTool>
122 166 </item>
123 167 <item path="build/morfeusz/default_fsa.cpp" ex="false" tool="1" flavor2="4">
124 168 <ccTool flags="1">
125   - <incDir>
126   - <pElem>build</pElem>
127   - <pElem>morfeusz</pElem>
128   - </incDir>
129 169 <preprocessorList>
130 170 <Elem>libmorfeusz_EXPORTS</Elem>
131 171 </preprocessorList>
... ... @@ -136,17 +176,13 @@
136 176 tool="1"
137 177 flavor2="4">
138 178 <ccTool flags="1">
139   - <incDir>
140   - <pElem>build</pElem>
141   - <pElem>morfeusz</pElem>
142   - </incDir>
143 179 <preprocessorList>
144 180 <Elem>libmorfeusz_EXPORTS</Elem>
145 181 </preprocessorList>
146 182 </ccTool>
147 183 </item>
148 184 <item path="build/morfeusz/java/swigJAVA.cpp" ex="false" tool="1" flavor2="4">
149   - <ccTool flags="3">
  185 + <ccTool flags="1">
150 186 </ccTool>
151 187 </item>
152 188 <item path="build/morfeusz/morfeuszJAVA_wrap.cxx"
... ... @@ -155,14 +191,11 @@
155 191 flavor2="8">
156 192 <ccTool>
157 193 <incDir>
158   - <pElem>build</pElem>
159   - <pElem>morfeusz</pElem>
160 194 <pElem>build/morfeusz</pElem>
161 195 <pElem>/usr/lib/jvm/default-java/include</pElem>
162 196 <pElem>build/morfeusz/java</pElem>
163 197 </incDir>
164 198 <preprocessorList>
165   - <Elem>NDEBUG</Elem>
166 199 <Elem>_OPTIMIZE__=1</Elem>
167 200 <Elem>__PIC__=2</Elem>
168 201 <Elem>__pic__=2</Elem>
... ... @@ -181,8 +214,6 @@
181 214 flavor2="4">
182 215 <ccTool flags="1">
183 216 <incDir>
184   - <pElem>build</pElem>
185   - <pElem>morfeusz</pElem>
186 217 <pElem>/usr/lib/perl/5.14/CORE</pElem>
187 218 <pElem>build/morfeusz/perl</pElem>
188 219 </incDir>
... ... @@ -197,14 +228,11 @@
197 228 flavor2="8">
198 229 <ccTool>
199 230 <incDir>
200   - <pElem>build</pElem>
201   - <pElem>morfeusz</pElem>
202 231 <pElem>build/morfeusz</pElem>
203 232 <pElem>/usr/include/python2.7</pElem>
204 233 <pElem>build/morfeusz/python</pElem>
205 234 </incDir>
206 235 <preprocessorList>
207   - <Elem>NDEBUG</Elem>
208 236 <Elem>_OPTIMIZE__=1</Elem>
209 237 <Elem>__PIC__=2</Elem>
210 238 <Elem>__pic__=2</Elem>
... ... @@ -218,22 +246,40 @@
218 246 </ccTool>
219 247 </item>
220 248 <item path="build/morfeusz/python/swigPYTHON.cpp"
  249 + ex="true"
  250 + tool="3"
  251 + flavor2="4">
  252 + </item>
  253 + <item path="build/morfeusz/wrappers/java/swigJAVA.cpp"
221 254 ex="false"
222 255 tool="1"
223 256 flavor2="4">
  257 + <ccTool flags="1">
  258 + </ccTool>
  259 + </item>
  260 + <item path="build/morfeusz/wrappers/morfeuszPERL_wrap.cxx"
  261 + ex="false"
  262 + tool="1"
  263 + flavor2="4">
  264 + <ccTool flags="1">
  265 + <incDir>
  266 + <pElem>/usr/lib/perl/5.14/CORE</pElem>
  267 + <pElem>build/morfeusz/wrappers/perl</pElem>
  268 + </incDir>
  269 + <preprocessorList>
  270 + <Elem>morfeusz_perl_EXPORTS</Elem>
  271 + </preprocessorList>
  272 + </ccTool>
224 273 </item>
225 274 <item path="build1/morfeusz/java/swigJAVA.cpp" ex="false" tool="1" flavor2="4">
226 275 </item>
227 276 <item path="default_fsa.cpp" ex="false" tool="1" flavor2="4">
228 277 <ccTool>
229 278 <incDir>
230   - <pElem>build</pElem>
231   - <pElem>morfeusz</pElem>
232 279 <pElem>build/morfeusz</pElem>
233 280 <pElem>morfeusz/build/morfeusz</pElem>
234 281 </incDir>
235 282 <preprocessorList>
236   - <Elem>NDEBUG</Elem>
237 283 <Elem>libmorfeusz_EXPORTS</Elem>
238 284 </preprocessorList>
239 285 </ccTool>
... ... @@ -241,13 +287,10 @@
241 287 <item path="default_synth_fsa.cpp" ex="false" tool="1" flavor2="4">
242 288 <ccTool>
243 289 <incDir>
244   - <pElem>build</pElem>
245   - <pElem>morfeusz</pElem>
246 290 <pElem>build/morfeusz</pElem>
247 291 <pElem>morfeusz/build/morfeusz</pElem>
248 292 </incDir>
249 293 <preprocessorList>
250   - <Elem>NDEBUG</Elem>
251 294 <Elem>libmorfeusz_EXPORTS</Elem>
252 295 </preprocessorList>
253 296 </ccTool>
... ... @@ -255,34 +298,32 @@
255 298 <folder path="0">
256 299 <ccTool>
257 300 <incDir>
258   - <pElem>build</pElem>
259   - <pElem>morfeusz</pElem>
260 301 <pElem>build/morfeusz</pElem>
261 302 </incDir>
262 303 </ccTool>
263 304 </folder>
264   - <folder path="0/charset">
  305 + <folder path="0/case">
265 306 <ccTool>
266 307 <preprocessorList>
267 308 <Elem>libmorfeusz_EXPORTS</Elem>
268 309 </preprocessorList>
269 310 </ccTool>
270 311 </folder>
271   - <folder path="0/cli">
  312 + <folder path="0/charset">
272 313 <ccTool>
273 314 <preprocessorList>
274 315 <Elem>libmorfeusz_EXPORTS</Elem>
275 316 </preprocessorList>
276 317 </ccTool>
277 318 </folder>
278   - <folder path="0/decoder">
  319 + <folder path="0/cli">
279 320 <ccTool>
280 321 <preprocessorList>
281 322 <Elem>libmorfeusz_EXPORTS</Elem>
282 323 </preprocessorList>
283 324 </ccTool>
284 325 </folder>
285   - <folder path="0/deserializer">
  326 + <folder path="0/deserialization">
286 327 <ccTool>
287 328 <preprocessorList>
288 329 <Elem>libmorfeusz_EXPORTS</Elem>
... ... @@ -303,35 +344,24 @@
303 344 </preprocessorList>
304 345 </ccTool>
305 346 </folder>
306   - <folder path="build">
  347 + <folder path="build/morfeusz/wrappers/java">
307 348 <ccTool>
308 349 <incDir>
309   - <pElem>build</pElem>
310   - <pElem>morfeusz</pElem>
311   - <pElem>build/morfeusz</pElem>
  350 + <pElem>/usr/lib/jvm/default-java/include</pElem>
312 351 </incDir>
313 352 <preprocessorList>
314   - <Elem>NDEBUG</Elem>
315   - <Elem>__PIC__=2</Elem>
316   - <Elem>__pic__=2</Elem>
317   - <Elem>libmorfeusz_EXPORTS</Elem>
  353 + <Elem>libjmorfeusz_EXPORTS</Elem>
318 354 </preprocessorList>
319   - <undefinedList>
320   - <Elem>__GCC_HAVE_DWARF2_CFI_ASM=1</Elem>
321   - </undefinedList>
322 355 </ccTool>
323 356 </folder>
324 357 <folder path="java">
325 358 <ccTool>
326 359 <incDir>
327   - <pElem>build</pElem>
328   - <pElem>morfeusz</pElem>
329 360 <pElem>build/morfeusz</pElem>
330 361 <pElem>build1</pElem>
331 362 <pElem>/usr/lib/jvm/java-6-openjdk/include</pElem>
332 363 </incDir>
333 364 <preprocessorList>
334   - <Elem>NDEBUG</Elem>
335 365 <Elem>libjmorfeusz_EXPORTS</Elem>
336 366 <Elem>libmorfeusz_EXPORTS</Elem>
337 367 </preprocessorList>
... ... @@ -340,12 +370,9 @@
340 370 <folder path="morfeusz/java">
341 371 <ccTool>
342 372 <incDir>
343   - <pElem>build</pElem>
344   - <pElem>morfeusz</pElem>
345   - <pElem>/usr/lib/jvm/java-6-openjdk/include</pElem>
  373 + <pElem>/usr/lib/jvm/default-java/include</pElem>
346 374 </incDir>
347 375 <preprocessorList>
348   - <Elem>NDEBUG</Elem>
349 376 <Elem>libjmorfeusz_EXPORTS</Elem>
350 377 </preprocessorList>
351 378 </ccTool>
... ... @@ -353,21 +380,10 @@
353 380 <folder path="morfeusz/python">
354 381 <ccTool>
355 382 <incDir>
356   - <pElem>morfeusz</pElem>
357 383 <pElem>/usr/include/python2.7</pElem>
358 384 </incDir>
359   - <preprocessorList>
360   - <Elem>NDEBUG</Elem>
361   - </preprocessorList>
362 385 </ccTool>
363 386 </folder>
364   - <item path="morfeusz/CasePatternHelper.cpp" ex="false" tool="1" flavor2="4">
365   - <ccTool flags="1">
366   - <preprocessorList>
367   - <Elem>libmorfeusz_EXPORTS</Elem>
368   - </preprocessorList>
369   - </ccTool>
370   - </item>
371 387 <item path="morfeusz/Environment.cpp" ex="false" tool="1" flavor2="4">
372 388 <ccTool flags="1">
373 389 <preprocessorList>
... ... @@ -383,17 +399,8 @@
383 399 </ccTool>
384 400 </item>
385 401 <item path="morfeusz/Morfeusz.cpp" ex="false" tool="1" flavor2="4">
386   - <ccTool flags="3">
387   - <preprocessorList>
388   - <Elem>NDEBUG</Elem>
389   - <Elem>libmorfeusz_EXPORTS</Elem>
390   - </preprocessorList>
391   - </ccTool>
392   - </item>
393   - <item path="morfeusz/MorphDeserializer.cpp" ex="false" tool="1" flavor2="4">
394   - <ccTool flags="3">
  402 + <ccTool flags="2">
395 403 <preprocessorList>
396   - <Elem>NDEBUG</Elem>
397 404 <Elem>libmorfeusz_EXPORTS</Elem>
398 405 </preprocessorList>
399 406 </ccTool>
... ... @@ -419,36 +426,36 @@
419 426 </preprocessorList>
420 427 </ccTool>
421 428 </item>
422   - <item path="morfeusz/charset/CaseConverter.cpp" ex="false" tool="1" flavor2="4">
  429 + <item path="morfeusz/case/CaseConverter.cpp" ex="false" tool="1" flavor2="4">
423 430 <ccTool flags="1">
424 431 </ccTool>
425 432 </item>
426   - <item path="morfeusz/charset/CharsetConverter.cpp"
  433 + <item path="morfeusz/case/CasePatternHelper.cpp"
427 434 ex="false"
428 435 tool="1"
429 436 flavor2="4">
430   - <ccTool flags="3">
431   - <preprocessorList>
432   - <Elem>NDEBUG</Elem>
433   - </preprocessorList>
434   - </ccTool>
435   - </item>
436   - <item path="morfeusz/charset/TextReader.cpp" ex="false" tool="1" flavor2="4">
437 437 <ccTool flags="1">
438 438 </ccTool>
439 439 </item>
440   - <item path="morfeusz/charset/caseconv.cpp" ex="false" tool="1" flavor2="4">
  440 + <item path="morfeusz/case/caseconv.cpp" ex="false" tool="1" flavor2="4">
441 441 <ccTool flags="1">
442 442 </ccTool>
443 443 </item>
  444 + <item path="morfeusz/charset/CharsetConverter.cpp"
  445 + ex="false"
  446 + tool="1"
  447 + flavor2="4">
  448 + </item>
  449 + <item path="morfeusz/charset/TextReader.cpp" ex="false" tool="1" flavor2="4">
  450 + </item>
444 451 <item path="morfeusz/charset/conversion_tables.cpp"
445 452 ex="false"
446 453 tool="1"
447 454 flavor2="4">
448   - <ccTool flags="1">
449   - </ccTool>
450 455 </item>
451 456 <item path="morfeusz/cli/cli.cpp" ex="false" tool="1" flavor2="4">
  457 + <ccTool flags="1">
  458 + </ccTool>
452 459 </item>
453 460 <item path="morfeusz/const.cpp" ex="false" tool="1" flavor2="4">
454 461 <ccTool flags="1">
... ... @@ -457,29 +464,31 @@
457 464 </preprocessorList>
458 465 </ccTool>
459 466 </item>
460   - <item path="morfeusz/decoder/InterpretedChunksDecoder.cpp"
  467 + <item path="morfeusz/deserialization/InterpsGroupsReader.cpp"
461 468 ex="false"
462 469 tool="1"
463 470 flavor2="4">
464 471 </item>
465   - <item path="morfeusz/decoder/InterpretedChunksDecoder4Analyzer.cpp"
  472 + <item path="morfeusz/deserialization/MorphDeserializer.cpp"
466 473 ex="false"
467 474 tool="1"
468 475 flavor2="4">
469 476 </item>
470   - <item path="morfeusz/decoder/InterpretedChunksDecoder4Generator.cpp"
  477 + <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder.cpp"
471 478 ex="false"
472 479 tool="1"
473 480 flavor2="4">
  481 + <ccTool flags="1">
  482 + </ccTool>
474 483 </item>
475   - <item path="morfeusz/deserializer/InterpsGroupsReader.cpp"
  484 + <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Analyzer.cpp"
476 485 ex="false"
477 486 tool="1"
478 487 flavor2="4">
479 488 <ccTool flags="1">
480 489 </ccTool>
481 490 </item>
482   - <item path="morfeusz/deserializer/MorphDeserializer.cpp"
  491 + <item path="morfeusz/deserialization/morphInterps/InterpretedChunksDecoder4Generator.cpp"
483 492 ex="false"
484 493 tool="1"
485 494 flavor2="4">
... ... @@ -496,9 +505,6 @@
496 505 <incDir>
497 506 <pElem>build/fsa</pElem>
498 507 </incDir>
499   - <preprocessorList>
500   - <Elem>NDEBUG</Elem>
501   - </preprocessorList>
502 508 </ccTool>
503 509 </item>
504 510 <item path="morfeusz/fsa/test_recognize.cpp" ex="false" tool="1" flavor2="8">
... ... @@ -506,9 +512,6 @@
506 512 <incDir>
507 513 <pElem>build/fsa</pElem>
508 514 </incDir>
509   - <preprocessorList>
510   - <Elem>NDEBUG</Elem>
511   - </preprocessorList>
512 515 </ccTool>
513 516 </item>
514 517 <item path="morfeusz/fsa/test_speed.cpp" ex="false" tool="1" flavor2="8">
... ... @@ -516,43 +519,33 @@
516 519 <incDir>
517 520 <pElem>build/fsa</pElem>
518 521 </incDir>
519   - <preprocessorList>
520   - <Elem>NDEBUG</Elem>
521   - </preprocessorList>
522   - </ccTool>
523   - </item>
524   - <item path="morfeusz/main.cpp" ex="false" tool="1" flavor2="4">
525   - <ccTool>
526   - <preprocessorList>
527   - <Elem>NDEBUG</Elem>
528   - <Elem>_OPTIMIZE__=1</Elem>
529   - <Elem>libmorfeusz_EXPORTS</Elem>
530   - </preprocessorList>
531   - <undefinedList>
532   - <Elem>__NO_INLINE__</Elem>
533   - </undefinedList>
534 522 </ccTool>
535 523 </item>
536 524 <item path="morfeusz/morfeusz_analyzer.cpp" ex="false" tool="1" flavor2="4">
537   - <ccTool flags="2">
538   - <preprocessorList>
539   - <Elem>NDEBUG</Elem>
540   - </preprocessorList>
  525 + <ccTool flags="0">
541 526 </ccTool>
542 527 </item>
543 528 <item path="morfeusz/morfeusz_generator.cpp" ex="false" tool="1" flavor2="4">
  529 + <ccTool flags="0">
  530 + </ccTool>
544 531 </item>
545 532 <item path="morfeusz/segrules/SegrulesFSA.cpp" ex="false" tool="1" flavor2="4">
546   - <ccTool flags="1">
547   - </ccTool>
548 533 </item>
549 534 <item path="morfeusz/segrules/segrules.cpp" ex="false" tool="1" flavor2="4">
550   - <ccTool flags="1">
551   - </ccTool>
552 535 </item>
553   - <item path="morfeusz/test_recognize_dict.cpp" ex="false" tool="1" flavor2="4">
  536 + <item path="morfeusz/test/test_recognize_dict.cpp"
  537 + ex="false"
  538 + tool="1"
  539 + flavor2="4">
  540 + <ccTool flags="0">
  541 + </ccTool>
554 542 </item>
555   - <item path="morfeusz/test_result_equals.cpp" ex="false" tool="1" flavor2="4">
  543 + <item path="morfeusz/test/test_result_equals.cpp"
  544 + ex="false"
  545 + tool="1"
  546 + flavor2="4">
  547 + <ccTool flags="0">
  548 + </ccTool>
556 549 </item>
557 550 </conf>
558 551 </confs>
... ...
profile.sh
  View file @a65ee05
... ... @@ -4,9 +4,9 @@ rm -rf profbuild
4 4 mkdir -p profbuild
5 5 cd profbuild
6 6 cmake -D INPUT_DICTIONARIES=../input/dodatki.tab,../input/PoliMorfSmall.tab -D CMAKE_BUILD_TYPE=Debug -D CMAKE_CXX_FLAGS="-g -O2" -D CMAKE_SHARED_LINKER_FLAGS="-lprofiler" -D CMAKE_EXE_LINKER_FLAGS="-lprofiler" ..
7   -make -j4
  7 +make
8 8 rm -f /tmp/morfeusz.prof
9 9 export LD_PRELOAD="/usr/lib/libprofiler.so"
10 10 export CPUPROFILE="/tmp/morfeusz.prof"
11   -morfeusz/morfeusz_analyzer -i /tmp/dupadupa < /mnt/storage/morfeusz/sents30k > /dev/null
  11 +morfeusz/morfeusz_analyzer -i /home/wkieras/output/sgjp_analyzer.fsa < /mnt/storage/morfeusz/sents10k > /dev/null
12 12 ### pprof --gv profbuild/morfeusz/morfeusz_analyzer /tmp/morfeusz.prof
... ...
morfeusz/Toolchain-Darwin-amd64.cmake renamed to toolchains/Toolchain-Darwin-amd64.cmake
  View file @a65ee05
morfeusz/Toolchain-Linux-amd64.cmake renamed to toolchains/Toolchain-Linux-amd64.cmake
  View file @a65ee05
morfeusz/Toolchain-Linux-i386.cmake renamed to toolchains/Toolchain-Linux-i386.cmake
  View file @a65ee05
morfeusz/Toolchain-Windows-amd64.cmake renamed to toolchains/Toolchain-Windows-amd64.cmake
  View file @a65ee05
morfeusz/Toolchain-Windows-i386.cmake renamed to toolchains/Toolchain-Windows-i386.cmake
  View file @a65ee05