diff --git a/CMakeLists.txt b/CMakeLists.txt
index c71f8cc082c79f9255c2f0a2c1d56ae8c53e7643..952c9deab6259b08f73a09e4bdea9aaf9ac8172e 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -67,4 +67,5 @@ add_subdirectory(libpwrutils)
add_subdirectory(libcorpus2)
add_subdirectory(tagset-tool)
add_subdirectory(tests)
+# add_subdirectory(swig)
diff --git a/CMakeScripts/FindCorpus2.cmake b/CMakeScripts/FindCorpus2.cmake
new file mode 100644
index 0000000000000000000000000000000000000000..da710906b8c363ba5f23e3fd1e96408cb5ad38b6
--- /dev/null
+++ b/CMakeScripts/FindCorpus2.cmake
@@ -0,0 +1,49 @@
+IF (Corpus2_INCLUDE_DIR AND Corpus2_LIBRARY)
+#Already in cache
+ SET(Corpus2_FOUND TRUE)
+ELSE (Corpus2_INCLUDE_DIR AND Corpus2_LIBRARY)
+ FIND_PATH(Corpus2_INCLUDE_DIR libcorpus2/token.h /usr/include /usr/local/include)
+
+ FIND_LIBRARY(Corpus2_LIBRARY corpus2 /usr/lib /usr/local/lib)
+
+ MARK_AS_ADVANCED(Corpus2_LIBRARY)
+ MARK_AS_ADVANCED(Corpus2_INCLUDE_DIR)
+
+ IF (Corpus2_INCLUDE_DIR AND Corpus2_LIBRARY)
+ SET(Corpus2_FOUND TRUE)
+ ENDIF (Corpus2_INCLUDE_DIR AND Corpus2_LIBRARY)
+
+ENDIF (Corpus2_INCLUDE_DIR AND Corpus2_LIBRARY)
+
+
+IF (Corpus2_FOUND)
+ set(Corpus2_VERSION 0.0.0)
+ FIND_FILE(_Corpus2_VERSION_FILE libcorpus2/version.h ${Corpus2_INCLUDE_DIR})
+ MARK_AS_ADVANCED(_Corpus2_VERSION_FILE)
+ IF (_Corpus2_VERSION_FILE)
+ FILE(READ ${_Corpus2_VERSION_FILE} _Corpus2_VERSION_CONENTS)
+ STRING(REGEX REPLACE ".*#define LIBCORPUS2_VERSION \\\"([0-9.]+)\\\".*" "\\1" Corpus2_VERSION "${_Corpus2_VERSION_CONENTS}")
+ ENDIF (_Corpus2_VERSION_FILE)
+ IF (Corpus2_FIND_VERSION)
+ IF (Corpus2_VERSION VERSION_LESS Corpus2_FIND_VERSION)
+ IF (Corpus2_FIND_REQUIRED)
+ MESSAGE(${_Corpus2_VERSION_FILE})
+ MESSAGE(FATAL_ERROR "Corpus2 version too old: ${Corpus2_VERSION}, requested >= ${Corpus2_FIND_VERSION}")
+ ELSE (Corpus2_FIND_REQUIRED)
+ IF (NOT Corpus2_FIND_QUIETLY)
+ MESSAGE(STATUS "Corpus2 version too old: ${Corpus2_VERSION}, requested >= ${Corpus2_FIND_VERSION}")
+ ENDIF (NOT Corpus2_FIND_QUIETLY)
+ ENDIF (Corpus2_FIND_REQUIRED)
+ set(Corpus2_FOUND False)
+ ENDIF (Corpus2_VERSION VERSION_LESS Corpus2_FIND_VERSION)
+ ENDIF (Corpus2_FIND_VERSION)
+ IF (NOT Corpus2_FIND_QUIETLY)
+ MESSAGE(STATUS "Found libcorpus2 ${Corpus2_VERSION}: ${Corpus2_LIBRARY}")
+ ENDIF (NOT Corpus2_FIND_QUIETLY)
+ELSE (Corpus2_FOUND)
+ IF (Corpus2_FIND_REQUIRED)
+ MESSAGE(FATAL_ERROR "Could not find libcorpus2")
+ ELSE (Corpus2_FIND_REQUIRED)
+ MESSAGE(STATUS "libcorpus2 not found")
+ ENDIF (Corpus2_FIND_REQUIRED)
+ENDIF (Corpus2_FOUND)
diff --git a/CMakeScripts/FindPwrUtils.cmake b/CMakeScripts/FindPwrUtils.cmake
new file mode 100644
index 0000000000000000000000000000000000000000..858e922e5d0fb3daa4c0ba699a9674b42b9f9a38
--- /dev/null
+++ b/CMakeScripts/FindPwrUtils.cmake
@@ -0,0 +1,49 @@
+IF (PwrUtils_INCLUDE_DIR AND PwrUtils_LIBRARY)
+#already in cache
+ SET(PwrUtils_FOUND TRUE)
+ELSE (PwrUtils_INCLUDE_DIR AND PwrUtils_LIBRARY)
+
+ FIND_PATH(PwrUtils_INCLUDE_DIR libpwrutils/foreach.h /usr/include /usr/local/include )
+
+ FIND_LIBRARY(PwrUtils_LIBRARY pwrutils /usr/lib /usr/local/lib)
+
+ MARK_AS_ADVANCED(PwrUtils_LIBRARY)
+ MARK_AS_ADVANCED(PwrUtils_INCLUDE_DIR)
+
+ IF (PwrUtils_INCLUDE_DIR AND PwrUtils_LIBRARY)
+ SET(PwrUtils_FOUND TRUE)
+ ENDIF (PwrUtils_INCLUDE_DIR AND PwrUtils_LIBRARY)
+ENDIF (PwrUtils_INCLUDE_DIR AND PwrUtils_LIBRARY)
+
+
+IF (PwrUtils_FOUND)
+ set(PwrUtils_VERSION 0.0.0)
+ FIND_FILE(_PwrUtils_VERSION_FILE libpwrutils/version.h ${PwrUtils_INCLUDE_DIR})
+ MARK_AS_ADVANCED(_PwrUtils_VERSION_FILE)
+ IF (_PwrUtils_VERSION_FILE)
+ FILE(READ ${_PwrUtils_VERSION_FILE} _PwrUtils_VERSION_CONENTS)
+ STRING(REGEX REPLACE ".*#define LIBPWRUTILS_VERSION \\\"([0-9.]+)\\\".*" "\\1" PwrUtils_VERSION "${_PwrUtils_VERSION_CONENTS}")
+ ENDIF (_PwrUtils_VERSION_FILE)
+ IF (PwrUtils_FIND_VERSION)
+ IF (PwrUtils_VERSION VERSION_LESS PwrUtils_FIND_VERSION)
+ IF (PwrUtils_FIND_REQUIRED)
+ MESSAGE(${_PwrUtils_VERSION_FILE})
+ MESSAGE(FATAL_ERROR "PwrUtils version too old: ${PwrUtils_VERSION}, requested >= ${PwrUtils_FIND_VERSION}")
+ ELSE (PwrUtils_FIND_REQUIRED)
+ IF (NOT PwrUtils_FIND_QUIETLY)
+ MESSAGE(STATUS "PwrUtils version too old: ${PwrUtils_VERSION}, requested >= ${PwrUtils_FIND_VERSION}")
+ ENDIF (NOT PwrUtils_FIND_QUIETLY)
+ ENDIF (PwrUtils_FIND_REQUIRED)
+ set(PwrUtils_FOUND False)
+ ENDIF (PwrUtils_VERSION VERSION_LESS PwrUtils_FIND_VERSION)
+ ENDIF (PwrUtils_FIND_VERSION)
+ IF (NOT PwrUtils_FIND_QUIETLY)
+ MESSAGE(STATUS "Found libpwrutils ${PwrUtils_VERSION}: ${PwrUtils_LIBRARY}")
+ ENDIF (NOT PwrUtils_FIND_QUIETLY)
+ELSE (PwrUtils_FOUND)
+ IF (PwrUtils_FIND_REQUIRED)
+ MESSAGE(FATAL_ERROR "Could not find libpwrutils")
+ ELSE (PwrUtils_FIND_REQUIRED)
+ MESSAGE(STATUS "libpwrutils not found")
+ ENDIF (PwrUtils_FIND_REQUIRED)
+ENDIF (PwrUtils_FOUND)
diff --git a/corpus2data/simple.tagset b/corpus2data/simple.tagset
new file mode 100644
index 0000000000000000000000000000000000000000..137d9c74c426465462323458d05bc4ff0d19fb79
--- /dev/null
+++ b/corpus2data/simple.tagset
@@ -0,0 +1,14 @@
+[ATTR]
+
+[POS]
+noun
+adj
+adv
+verb
+qub
+other
+interp
+unk
+
+[IGN]
+unk
diff --git a/libcorpus2/io/cclreader.cpp b/libcorpus2/io/cclreader.cpp
index eadd580e7dac8c36dd7b7bff4203f96548d90de2..4559ed8e1c3ff8258ca257540a6d612e86ab6a00 100644
--- a/libcorpus2/io/cclreader.cpp
+++ b/libcorpus2/io/cclreader.cpp
@@ -82,7 +82,7 @@ CclReader::CclReader(const Tagset& tagset, const std::string& filename, bool dis
{
this->is_owned_.reset(new std::ifstream(filename.c_str(), std::ifstream::in));
- if (this->is_owned_->bad()) {
+ if (!this->is_owned_->good()) {
throw Corpus2Error("File not found!");
}
else {
diff --git a/libcorpus2/io/fastxces.cpp b/libcorpus2/io/fastxces.cpp
index a4813bbc53ef4af326eca28ae3368ba0ded5c1b0..6eaedaf8390214e4c48856b2cc8e054ccdbfa0dc 100644
--- a/libcorpus2/io/fastxces.cpp
+++ b/libcorpus2/io/fastxces.cpp
@@ -175,7 +175,7 @@ FastXcesReader::FastXcesReader(const Tagset &tagset, const std::string &filename
{
this->is_owned_.reset(new std::ifstream(filename.c_str(), std::ifstream::in));
- if (this->is_owned_->bad()) {
+ if (!this->is_owned_->good()) {
throw Corpus2Error("File not found!");
}
else {
diff --git a/libcorpus2/io/rft.cpp b/libcorpus2/io/rft.cpp
index 73003a6b8911258a1015e095f3b8008c85738734..ead40217893a217749ec7e8b62a945c5dfdb36e6 100644
--- a/libcorpus2/io/rft.cpp
+++ b/libcorpus2/io/rft.cpp
@@ -116,7 +116,7 @@ RftReader::RftReader(const Tagset& tagset, const std::string& filename, bool dis
, mbt_dialect_(mbt_dialect)
{
is_owned_.reset(new std::ifstream(filename.c_str(), std::ifstream::in));
- if (this->is_owned_->bad()) {
+ if (!this->is_owned_->good()) {
throw Corpus2Error("File not found!");
}
else {
diff --git a/libcorpus2/io/xcesreader.cpp b/libcorpus2/io/xcesreader.cpp
index bf14d9dc0080a6bc1a0f2a691d834169a0c7a9d1..6db3428d55e407ed64e14112240e747ec180059a 100644
--- a/libcorpus2/io/xcesreader.cpp
+++ b/libcorpus2/io/xcesreader.cpp
@@ -52,8 +52,7 @@ XcesReader::XcesReader(const Tagset& tagset, const std::string& filename, bool d
impl_(new XcesReaderImpl(*this, chunk_buf_, disamb_only, disamb_sh))
{
this->is_owned_.reset(new std::ifstream(filename.c_str(), std::ifstream::in));
-
- if (this->is_owned_->bad()) {
+ if (!this->is_owned_->good()) {
throw Corpus2Error("File not found!");
}
else {
diff --git a/libcorpus2/io/xceswriter.cpp b/libcorpus2/io/xceswriter.cpp
index 33693ce8dc8ee9cc51dd7de1c7cc693cb6e7ecfc..bf1cd78d9da558270cc0a2d0b8fd4dc217d47fcb 100644
--- a/libcorpus2/io/xceswriter.cpp
+++ b/libcorpus2/io/xceswriter.cpp
@@ -71,7 +71,10 @@ void XcesWriter::write_chunk(const Chunk &c)
}
write_sentence(*s);
}
- if (use_indent_) indent_less();
+ if (new_chunk) { //no sentences in chunk
+ paragraph_head(c);
+ }
+ if (use_indent_ && !new_chunk) indent_less();
osi() << "</chunk>\n";
}
diff --git a/libcorpus2/lexeme.cpp b/libcorpus2/lexeme.cpp
index be0f8aeb7c82a1343c7b6366ff3f96affbbcc718..d5370ed662ee7974505ec9732c904cf6377e0fc6 100644
--- a/libcorpus2/lexeme.cpp
+++ b/libcorpus2/lexeme.cpp
@@ -33,6 +33,12 @@ Lexeme Lexeme::create(const UnicodeString& lemma, const Tag& tag)
return Lexeme(lemma, tag);
}
+Lexeme Lexeme::create_utf8(const std::string& lemma_utf8, const Tag& tag)
+{
+ return Lexeme(UnicodeString::fromUTF8(lemma_utf8), tag);
+}
+
+
bool Lexeme::is_null() const
{
return lemma().length() == 0 || tag().is_null();
diff --git a/libcorpus2/lexeme.h b/libcorpus2/lexeme.h
index fbf82a0a8957e06ea7dab83fb61f27c6a351ca2e..4c0119ab58d91c9b932be2eba46cf4373272040b 100644
--- a/libcorpus2/lexeme.h
+++ b/libcorpus2/lexeme.h
@@ -53,6 +53,9 @@ public:
/// Helper creation function
static Lexeme create(const UnicodeString& lemma, const Tag& tag);
+ /// Helper creation function, UTF-8 variant
+ static Lexeme create_utf8(const std::string& lemma_utf8, const Tag& tag);
+
/// Lemma accessor
const UnicodeString& lemma() const {
return lemma_;
@@ -63,6 +66,11 @@ public:
lemma_ = l;
}
+ /// Lemma setter, UTF-8 variant
+ void set_lemma_utf8(const std::string& lemma_utf8) {
+ lemma_ = UnicodeString::fromUTF8(lemma_utf8);
+ }
+
/// UTF-8 lemma convenience accessor
const std::string lemma_utf8() const {
return PwrNlp::to_utf8(lemma_);
diff --git a/libcorpus2/token.cpp b/libcorpus2/token.cpp
index dba032254ca1c2e122be2429cd41d4c5ae776a86..38f3bdb2c4f92ed92691fe33662ac9bfc979dce2 100644
--- a/libcorpus2/token.cpp
+++ b/libcorpus2/token.cpp
@@ -32,6 +32,12 @@ Token::Token(const UnicodeString &orth, PwrNlp::Whitespace::Enum wa)
{
}
+Token* Token::create_utf8(const std::string& orth_utf8,
+ PwrNlp::Whitespace::Enum wa /*= PwrNlp::Whitespace::Space*/)
+{
+ return new Token(UnicodeString::fromUTF8(orth_utf8), wa);
+}
+
Token* Token::clone() const
{
Token* t = new Token();
diff --git a/libcorpus2/token.h b/libcorpus2/token.h
index 13c623fb8895955913de658f8e87225a336cf190..ab0992271655081f2c4d0f8442df0ec4f4c3bd07 100644
--- a/libcorpus2/token.h
+++ b/libcorpus2/token.h
@@ -54,6 +54,10 @@ public:
/// Create a Token with the given orth and whitespace amount
Token(const UnicodeString& orth, PwrNlp::Whitespace::Enum wa);
+ /// Token creation, UTF-8
+ static Token* create_utf8(const std::string& orth_utf8,
+ PwrNlp::Whitespace::Enum wa = PwrNlp::Whitespace::Space);
+
/// Create a duplicate Token
Token* clone() const;
@@ -72,6 +76,11 @@ public:
orth_ = orth;
}
+ /// Orth setter (UTF-8)
+ void set_orth_utf8(const std::string& orth_utf8) {
+ orth_ = UnicodeString::fromUTF8(orth_utf8);
+ }
+
/// WA getter
const PwrNlp::Whitespace::Enum& wa() const {
return wa_;
diff --git a/libpwrutils/pathsearch.cpp b/libpwrutils/pathsearch.cpp
index c89b7a4bf1af495ca866ffbc62bb7a3656d24ff3..1584d79a14af76d3011954dbd30ca505260a00c9 100644
--- a/libpwrutils/pathsearch.cpp
+++ b/libpwrutils/pathsearch.cpp
@@ -72,13 +72,24 @@ const std::string& PathSearcherBase::get_path_separator() const
}
std::string PathSearcherBase::find_file(const std::string& filename,
- const std::string& info)
+ const std::string& info) const
{
boost::filesystem::path i(filename);
+ if (i.is_complete()) {
+ if (boost::filesystem::exists(i) &&
+ !boost::filesystem::is_directory(i)) {
+ if (verbose_loading_) {
+ std::cerr << "Found " << info << " file: "
+ << i.string() << "\n";
+ }
+ return i.string();
+ }
+ return "";
+ }
foreach (const std::string& s, paths_) {
boost::filesystem::path pi = s / i;
if (boost::filesystem::exists(pi) &&
- boost::filesystem::is_regular(pi)) {
+ !boost::filesystem::is_directory(pi)) {
if (verbose_loading_) {
std::cerr << "Found " << info << " file: "
<< pi.string() << "\n";
@@ -90,7 +101,7 @@ std::string PathSearcherBase::find_file(const std::string& filename,
}
bool PathSearcherBase::open_stream(const std::string& filename,
- std::ifstream& ifs, const std::string& info)
+ std::ifstream& ifs, const std::string& info) const
{
std::string f = find_file(filename, info);
if (!f.empty()) {
diff --git a/libpwrutils/pathsearch.h b/libpwrutils/pathsearch.h
index 956f68deab9b0282b4b33d5a72c408495a9626a0..c33aaa7e677b5d84bd2f12bfeb71295b90a5b4a7 100644
--- a/libpwrutils/pathsearch.h
+++ b/libpwrutils/pathsearch.h
@@ -73,7 +73,7 @@ public:
* is on. Empty info string suppreses loading info.
*/
std::string find_file(const std::string& filename,
- const std::string& info = "");
+ const std::string& info = "") const;
/**
* Open a file stream for a file in the library search path
@@ -83,7 +83,22 @@ public:
* is on. Empty info string suppreses loading info.
*/
bool open_stream(const std::string& filename, std::ifstream& ifs,
- const std::string& info = "");
+ const std::string& info = "") const;
+
+ /**
+ * Convenience wrapper around find_file to throw an exception
+ * when the file is not found.
+ * Virtual, as it throws the exception defined by the child class.
+ */
+ virtual std::string find_file_or_throw(const std::string& filename,
+ const std::string& where) const = 0;
+
+ /**
+ * Wrapper around open_stream to throw an exception when the file is not
+ * found. Virtual, as it throws the exception defined by the child class.
+ */
+ virtual void open_stream_or_throw(const std::string& filename,
+ std::ifstream& ifs, const std::string& where) const = 0;
/**
* Look for files matching a condition.
@@ -122,14 +137,14 @@ public:
* when the file is not found.
*/
std::string find_file_or_throw(const std::string& filename,
- const std::string& where);
+ const std::string& where) const;
/**
* Convenience template wrapper around open_stream to throw an
* exception when the file is not found.
*/
void open_stream_or_throw(const std::string& filename,
- std::ifstream& ifs, const std::string& where);
+ std::ifstream& ifs, const std::string& where) const;
};
@@ -158,7 +173,7 @@ private:
template<class E>
std::string PathSearcher<E>::find_file_or_throw(
- const std::string& filename, const std::string& info)
+ const std::string& filename, const std::string& info) const
{
std::string fn = find_file(filename, info);
if (fn.empty()) {
@@ -169,7 +184,7 @@ std::string PathSearcher<E>::find_file_or_throw(
template<class E>
void PathSearcher<E>::open_stream_or_throw(const std::string& filename,
- std::ifstream& ifs, const std::string& info)
+ std::ifstream& ifs, const std::string& info) const
{
if (!open_stream(filename, ifs, info)) {
throw E(filename, get_search_path_string(), info);
diff --git a/swig/CMakeLists.txt b/swig/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..978ece174dff98284913c1d162a075b38422948d
--- /dev/null
+++ b/swig/CMakeLists.txt
@@ -0,0 +1,108 @@
+########## corpus2 wrappers -- SWIG ###############
+
+PROJECT(corpus2SwigWrap)
+
+find_package(Corpus2 1.0.8 REQUIRED)
+set(CORPUS2_LIBS ${Corpus2_LIBRARY})
+
+find_package(PwrUtils 1.0.1 REQUIRED)
+set(PWRUTILS_LIBS ${PwrUtils_LIBRARY})
+set(CORPUS2_PWR_LIBS ${CORPUS2_LIBS} ${PWRUTILS_LIBS})
+
+include_directories (${Libcorpus2_SOURCE_DIR} "../libcorpus2")
+include_directories (${Libpwrutils_SOURCE_DIR} "../libpwrutils")
+
+link_directories(${Libcorpus2_BINARY_DIR})
+
+FIND_PACKAGE(SWIG REQUIRED)
+INCLUDE(${SWIG_USE_FILE})
+
+find_package(PythonLibs)
+find_package(PythonInterp)
+# idea taken from pyplot build system
+execute_process(
+ COMMAND
+ ${PYTHON_EXECUTABLE} -c "from distutils import sysconfig; print sysconfig.get_python_lib(1,0,prefix='${CMAKE_INSTALL_EXEC_PREFIX}')"
+ OUTPUT_VARIABLE PYTHON_INSTDIR
+ OUTPUT_STRIP_TRAILING_WHITESPACE
+)
+message(STATUS "INFO: " "python lib: ${PYTHON_INSTDIR}" )
+
+INCLUDE_DIRECTORIES(${PYTHON_INCLUDE_PATH})
+INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR})
+
+# -----------------------------------------------------------------------------
+# -----------------------------------------------------------------------------
+
+SET(CMAKE_SWIG_FLAGS "")
+SET(SWIG_SRC_DIR ${SWIG_DIR}/libcorpus)
+
+# -----------------------------------------------------------------------------
+# corous2
+SET_SOURCE_FILES_PROPERTIES(corpus2.i PROPERTIES CPLUSPLUS ON)
+SET_SOURCE_FILES_PROPERTIES(corpus2.i PROPERTIES SWIG_FLAGS "-includeall" )
+SWIG_ADD_MODULE(corpus2 python corpus2.i )
+SWIG_LINK_LIBRARIES(corpus2 ${PYTHON_LIBRARIES} ${PWRUTILS_LIBS})
+SWIG_LINK_LIBRARIES(corpus2 ${PYTHON_LIBRARIES} ${CORPUS2_PWR_LIBS})
+
+# -----------------------------------------------------------------------------
+# -----------------------------------------------------------------------------
+
+set_target_properties(
+ _corpus2
+ PROPERTIES
+ INSTALL_NAME_DIR "${PYTHON_INSTDIR}"
+)
+
+set(PERM_SCRIPTS
+ OWNER_READ
+ OWNER_WRITE
+ OWNER_EXECUTE
+ GROUP_READ
+ GROUP_EXECUTE
+ WORLD_READ
+ WORLD_EXECUTE
+)
+
+# -----------------------------------------------------------------------------
+# Install python modules
+# -----------------------------------------------------------------------------
+
+install(
+ TARGETS _corpus2
+ LIBRARY
+ DESTINATION ${PYTHON_INSTDIR}
+ PERMISSIONS ${PERM_SCRIPTS}
+)
+
+install(
+ FILES ${CMAKE_CURRENT_BINARY_DIR}/corpus2.py
+ DESTINATION ${PYTHON_INSTDIR}
+ PERMISSIONS ${PERM_SCRIPTS}
+)
+
+# -----------------------------------------------------------------------------
+# Install SWIG files
+# -----------------------------------------------------------------------------
+
+install(
+ FILES libpwrnlperror.i
+ libcorpus2exception.i
+ libcorpus2exception.i
+ libcorpustag.i
+ libcorpustagset.i
+ libcorpustagsetmanager.i
+ libcorpuslexeme.i
+ libcorpustoken.i
+ libcorpussentence.i
+ libcorpuschunk.i
+ libcorpustokenwriter.i
+ libcorpustokenreader.i
+ libcorpusiob.i
+ libcorpusannotationchannel.i
+ libcorpusannotatedsentence.i
+ libcorpusannotationview.i
+ corpus2.i
+ DESTINATION ${SWIG_SRC_DIR}
+ PERMISSIONS ${PERM_SCRIPTS}
+)
diff --git a/swig/Makefile b/swig/Makefile
index 8ca3262b87b9911c64a79f8c4059738dd8de8cbc..9d34763f60a97d79b7445df5e8a06a6ba461b16f 100644
--- a/swig/Makefile
+++ b/swig/Makefile
@@ -13,7 +13,9 @@ ANTLRLIB=/usr/lib/libantlr-pic.a
CPPFLAGS=-fPIC -O2
-CBIN=libcorpustag.o \
+CBIN=libpwrnlperror.o \
+ libcorpus2exception.o \
+ libcorpustag.o \
libcorpustagset.o \
libcorpustagsetmanager.o \
libcorpuslexeme.o \
@@ -25,9 +27,12 @@ CBIN=libcorpustag.o \
libcorpusiob.o \
libcorpusannotationchannel.o \
libcorpusannotatedsentence.o \
- libcorpusannotationview.o
+ libcorpusannotationview.o \
+ corpus2.o
-CBINOUT=_boost_shared_ptr.so \
+CBINOUT=_libpwrnlperror.so \
+ _libcorpus2exception.so \
+ _boost_shared_ptr.so \
_libcorpustag.so \
_libcorpustagset.so \
_libcorpustagsetmanager.so \
@@ -40,9 +45,12 @@ CBINOUT=_boost_shared_ptr.so \
_libcorpusiob.so \
_libcorpusannotationchannel.so \
_libcorpusannotatedsentence.so \
- _libcorpusannotationview.so
+ _libcorpusannotationview.so \
+ _corpus2.so
-CWRAP=boost_shared_ptr_wrap.cxx \
+CWRAP=libpwrnlperror_wrap.cxx \
+ libcorpus2exception_wrap.cxx \
+ boost_shared_ptr_wrap.cxx \
libcorpustag_wrap.cxx \
libcorpustagset_wrap.cxx \
libcorpustagsetmanager_wrap.cxx \
@@ -55,9 +63,12 @@ CWRAP=boost_shared_ptr_wrap.cxx \
libcorpusiob_wrap.cxx \
libcorpusannotationchannel_wrap.cxx \
libcorpusannotatedsentence_wrap.cxx \
- libcorpusannotationview_wrap.cxx
+ libcorpusannotationview_wrap.cxx \
+ corpus2_wrap.cxx
-CWRAPBIN=boost_shared_ptr_wrap.o \
+CWRAPBIN=libpwrnlperror_wrap.o \
+ libcorpus2exception_wrap.o \
+ boost_shared_ptr_wrap.o \
libcorpustag_wrap.o \
libcorpustagset_wrap.o \
libcorpustagsetmanager_wrap.o \
@@ -70,9 +81,12 @@ CWRAPBIN=boost_shared_ptr_wrap.o \
libcorpusiob_wrap.o \
libcorpusannotationchannel_wrap.o \
libcorpusannotatedsentence_wrap.o \
- libcorpusannotationview_wrap.o
+ libcorpusannotationview_wrap.o \
+ corpus2_wrap.o
-PYMODULES=boost_shared_ptr.py \
+PYMODULES=libpwrnlperror.py \
+ libcorpus2exception.py \
+ boost_shared_ptr.py \
libcorpustag.py \
libcorpustagset.py \
libcorpustagsetmanager.py \
@@ -85,9 +99,12 @@ PYMODULES=boost_shared_ptr.py \
libcorpusiob.py \
libcorpusannotationchannel.py \
libcorpusannotatedsentence.py \
- libcorpusannotationview.py
+ libcorpusannotationview.py \
+ corpus2.py
-PYCBIN=boost_shared_ptr.pyc \
+PYCBIN=libpwrnlperror.pyc \
+ libcorpus2exception.pyc \
+ boost_shared_ptr.pyc \
libcorpustag.pyc \
libcorpustagset.pyc \
libcorpustagsetmanager.pyc \
@@ -100,7 +117,8 @@ PYCBIN=boost_shared_ptr.pyc \
libcorpusiob.pyc \
libcorpusannotationchannel.pyc \
libcorpusannotatedsentence.pyc \
- libcorpusannotationview.pyc
+ libcorpusannotationview.pyc \
+ corpus2.pyc
# -----------------------------------------------------------------------------
all:boost_shared_ptr.o $(CBIN)
@@ -116,10 +134,25 @@ boost_shared_ptr.o:
$(CPP) -c boost_shared_ptr_wrap.cxx -I$(PYTHONDIR) $(CPPFLAGS)
$(CPP) -shared boost_shared_ptr_wrap.o -o _boost_shared_ptr.so
+# -----------------------------------------------------------------------------
+# PwrNlpError wprapper
+# -----------------------------------------------------------------------------
+# PwrNlpError
+libpwrnlperror.o:
+ $(SWIG) $(SWIGOPTS_LANG) libpwrnlperror.i
+ $(CPP) -c libpwrnlperror_wrap.cxx -I$(PYTHONDIR) $(CPPFLAGS)
+ $(CPP) -shared libpwrnlperror_wrap.o -o _libpwrnlperror.so
# -----------------------------------------------------------------------------
# Corpus2 Wrappers
# -----------------------------------------------------------------------------
+
+libcorpus2exception.o:
+ $(SWIG) $(SWIGOPTS_LANG) libcorpus2exception.i
+ $(CPP) -c libcorpus2exception_wrap.cxx -I$(PYTHONDIR) $(CPPFLAGS)
+ $(CPP) -shared libcorpus2exception_wrap.o \
+ $(PWRUTILBIN) $(CORPUS2BIN) -o _libcorpus2exception.so
+
# Tag
libcorpustag.o:
$(SWIG) $(SWIGOPTS_LANG) libcorpustag.i
@@ -211,6 +244,15 @@ libcorpusannotationview.o:
$(CPP) -shared libcorpusannotationview_wrap.o \
$(PWRUTILBIN) $(CORPUS2BIN) -o _libcorpusannotationview.so
+# -----------------------------------------------------------------------------
+
+# Corpus2
+corpus2.o:
+ $(SWIG) $(SWIGOPTS_LANG) corpus2.i
+ $(CPP) -c corpus2_wrap.cxx -I$(PYTHONDIR) $(CPPFLAGS)
+ $(CPP) -shared corpus2_wrap.o \
+ $(PWRUTILBIN) $(CORPUS2BIN) -o _corpus2.so
+
# -----------------------------------------------------------------------------
clean:
rm -f $(CBIN) $(CBINOUT) $(CWRAP) $(CWRAPBIN) $(PYMODULES) $(PYCBIN)
diff --git a/swig/corpus2.i b/swig/corpus2.i
new file mode 100644
index 0000000000000000000000000000000000000000..a805a328edbf38ce4cf86f21915a120b9e2d5b60
--- /dev/null
+++ b/swig/corpus2.i
@@ -0,0 +1,23 @@
+#ifndef SWIG_CORPUS2_I
+#define SWIG_CORPUS2_I
+
+%module corpus2
+
+%include "boost_shared_ptr.i"
+%include "libcorpus2exception.i"
+%include "libcorpusannotatedsentence.i"
+%include "libcorpusannotationchannel.i"
+%include "libcorpusannotationview.i"
+%include "libcorpuschunk.i"
+%include "libcorpusiob.i"
+%include "libcorpuslexeme.i"
+%include "libcorpussentence.i"
+%include "libcorpustag.i"
+%include "libcorpustagset.i"
+%include "libcorpustagsetmanager.i"
+%include "libcorpustoken.i"
+%include "libcorpustokenreader.i"
+%include "libcorpustokenwriter.i"
+%include "libpwrnlperror.i"
+
+#endif /* SWIG_CORPUS2_I */
diff --git a/swig/libcorpus2exception.i b/swig/libcorpus2exception.i
new file mode 100644
index 0000000000000000000000000000000000000000..2ee4463ae619ffa4dd80c12e1aec8161682c2369
--- /dev/null
+++ b/swig/libcorpus2exception.i
@@ -0,0 +1,39 @@
+#ifndef SWIG_LIBCORPUS2_EXCEPTION_I
+#define SWIG_LIBCORPUS2_EXCEPTION_I
+
+%module libcorpus2exception
+%{
+ #include <libcorpus2/exception.h>
+%}
+
+%include "std_string.i"
+
+%include "libpwrnlperror.i"
+
+namespace Corpus2 {
+ class Corpus2Error : public PwrNlp::PwrNlpError {
+ public:
+ Corpus2Error(const std::string &what);
+ ~Corpus2Error() throw();
+
+ /* --------------------------------------------------------------------- */
+ std::string scope() const;
+ };
+
+ class FileNotFound : public Corpus2Error {
+ public:
+ FileNotFound(const std::string& filename, const std::string& paths,
+ const std::string& where);
+
+ ~FileNotFound() throw();
+
+ /* --------------------------------------------------------------------- */
+ std::string info() const;
+ std::string filename, paths, where;
+ };
+}
+
+using namespace std;
+using namespace Corpus2;
+
+#endif /* SWIG_LIBCORPUS2_EXCEPTION_I */
diff --git a/swig/libcorpusannotatedsentence.i b/swig/libcorpusannotatedsentence.i
index 0f85282b0414c9b45a9cd734ada6d38ffd926e4f..f7c27b109a0f30fdebc2fb2cfb3184d964abde29 100644
--- a/swig/libcorpusannotatedsentence.i
+++ b/swig/libcorpusannotatedsentence.i
@@ -8,6 +8,7 @@
%include "libcorpustoken.i"
%include "libcorpussentence.i"
+%include "libcorpus2exception.i"
%include "libcorpusannotationchannel.i"
%include "boost_shared_ptr.i"
@@ -21,6 +22,12 @@
%template(AnnotatedSentencePtr) boost::shared_ptr<Corpus2::AnnotatedSentence>;
namespace Corpus2 {
+ class MissingAnnotationChannel : public Corpus2Error {
+ public:
+ MissingAnnotationChannel(const std::string& name);
+ ~MissingAnnotationChannel() throw();
+ }; // MissingAnnotationChannel
+
class AnnotatedSentence : public Corpus2::Sentence {
public:
typedef std::map<std::string, AnnotationChannel> chan_map_t;
diff --git a/swig/libcorpuslexeme.i b/swig/libcorpuslexeme.i
index 058e8b7d449a9375f0f288543ec67670554413ee..e962845b00b107b7b538c3325819f2bfde3d95c4 100644
--- a/swig/libcorpuslexeme.i
+++ b/swig/libcorpuslexeme.i
@@ -24,13 +24,13 @@ namespace Corpus2 {
Lexeme(const UnicodeString& lemma, const Tag& tag);
static Lexeme create(const UnicodeString& lemma, const Tag& tag);
- // static Lexeme create_utf8(const std::string& lemma, const Tag& tag);
+ static Lexeme create_utf8(const std::string& lemma, const Tag& tag);
const UnicodeString& lemma() const;
const std::string lemma_utf8() const;
void set_lemma(const UnicodeString& l);
- // void set_lemma_utf8(const std::string& l);
+ void set_lemma_utf8(const std::string& l);
const Tag& tag() const;
void set_tag(const Tag& tag);
diff --git a/swig/libcorpussentence.i b/swig/libcorpussentence.i
index d580b81377394a74bc94a4e777b27d3158fea296..be08a46d7a9a32f0d8b04fa1f40c7d519e6601a4 100644
--- a/swig/libcorpussentence.i
+++ b/swig/libcorpussentence.i
@@ -47,6 +47,7 @@ namespace Corpus2 {
}
using namespace std;
+using namespace boost;
using namespace Corpus2;
#endif /* SWIG_LIBCORPUS2_SENTENCE_I */
diff --git a/swig/libcorpustagset.i b/swig/libcorpustagset.i
index 8030f6c23e5cd699dacb31b02dc1cd4a2fc5fb78..4b469a69295342a49ff34baec8f50bb2c6464735 100644
--- a/swig/libcorpustagset.i
+++ b/swig/libcorpustagset.i
@@ -7,6 +7,7 @@
%}
%include "libcorpustag.i"
+%include "libcorpus2exception.i"
%include "std_string.i"
%include "std_vector.i"
@@ -17,6 +18,36 @@
%template(StdStringVector) std::vector<std::string>;
namespace Corpus2 {
+ class Tagset;
+ class TagParseError : public Corpus2Error {
+ public:
+ TagParseError(const std::string &what, const std::string& val,
+ const std::string& tag, const std::string& tagset);
+
+ ~TagParseError() throw() {}
+
+ /* --------------------------------------------------------------------- */
+ std::string info() const;
+ std::string val, tag, tagset;
+ }; // TagParseError
+
+ class TagsetMismatch : public Corpus2Error {
+ public:
+ TagsetMismatch(const std::string& where, const Tagset& expected,
+ const Tagset& actual);
+
+ // TagsetMismatch(const std::string& where, tagset_idx_t expected,
+ // tagset_idx_t actual);
+
+ ~TagsetMismatch() throw() {}
+
+ /* --------------------------------------------------------------------- */
+ std::string info() const;
+
+ /* --------------------------------------------------------------------- */
+ // tagset_idx_t expected_id, actual_id;
+ }; // TagsetMismatch
+
class Tagset {
public:
diff --git a/swig/libcorpustagsetmanager.i b/swig/libcorpustagsetmanager.i
index 2e0ed4a82b6418c7d88a29c44f659df8c9a5040e..adc0b4c9aa55c8310043c97e3ab6d248c1b8eaac 100644
--- a/swig/libcorpustagsetmanager.i
+++ b/swig/libcorpustagsetmanager.i
@@ -7,6 +7,7 @@
%}
%include "libcorpustagset.i"
+%include "libcorpus2exception.i"
%include "std_string.i"
%include "std_vector.i"
@@ -16,6 +17,18 @@
%template(TagsetPtr) boost::shared_ptr<Tagset>;
namespace Corpus2 {
+ class TagsetNotFound : public Corpus2Error {
+ public:
+ // explicit TagsetNotFound(const tagset_idx_t id);
+ ~TagsetNotFound() throw() {}
+
+ /* --------------------------------------------------------------------- */
+ std::string info() const;
+
+ /* --------------------------------------------------------------------- */
+ // tagset_idx_t id;
+ };
+
class TagsetManager {
public:
TagsetManager();
@@ -28,6 +41,18 @@ namespace Corpus2 {
/* --------------------------------------------------------------------- */
};
+
+ %exception {
+ try {
+ $action
+ } catch (Corpus2::TagsetNotFound &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ } catch (PwrNlp::PwrNlpError &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ }
+ }
inline const Tagset& get_named_tagset(const std::string& name);
}
diff --git a/swig/libcorpustoken.i b/swig/libcorpustoken.i
index 079063b5d88d7c26eccbf75e83195f3f6833aff7..5d81bba2580a9fe4a3236f505c70345e51ad73f1 100644
--- a/swig/libcorpustoken.i
+++ b/swig/libcorpustoken.i
@@ -26,12 +26,14 @@ namespace Corpus2 {
Token();
Token(const UnicodeString& orth, PwrNlp::Whitespace::Enum wa);
Token* clone() const;
+
+ Token* create_utf8(const std::string& orth_utf8, PwrNlp::Whitespace::Enum wa = PwrNlp::Whitespace::Space);
const UnicodeString& orth() const;
std::string orth_utf8() const;
void set_orth(const UnicodeString& orth);
- // void set_orth_utf8(const std::string& orth);
+ void set_orth_utf8(const std::string& orth);
const PwrNlp::Whitespace::Enum& wa() const;
void set_wa(const PwrNlp::Whitespace::Enum& wa);
diff --git a/swig/libcorpustokenreader.i b/swig/libcorpustokenreader.i
index 935ba2159fcdfd9c23e1303ab6cfe52c6db738a2..93043ff8446bdb1adc53001c4da9781f2399dae1 100644
--- a/swig/libcorpustokenreader.i
+++ b/swig/libcorpustokenreader.i
@@ -30,11 +30,27 @@ namespace Corpus2 {
virtual ~TokenReader();
/* --------------------------------------------------------------------- */
+ %exception {
+ try {
+ $action
+ } catch (PwrNlp::PwrNlpError &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ }
+ }
static TokenReaderPtr create_path_reader(
const std::string& class_id,
const Tagset& tagset,
const std::string& path);
+ %exception {
+ try {
+ $action
+ } catch (PwrNlp::PwrNlpError &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ }
+ }
static TokenReaderPtr create_stream_reader(
const std::string& class_id,
const Tagset& tagset,
diff --git a/swig/libcorpustokenwriter.i b/swig/libcorpustokenwriter.i
index 19e914f1ab7c8dc68c9d17b971cfd6c88475a9f7..50ad969ab890b83f4fe22fc2ef5156bddce69932 100644
--- a/swig/libcorpustokenwriter.i
+++ b/swig/libcorpustokenwriter.i
@@ -39,11 +39,27 @@ namespace Corpus2 {
void finish();
/* --------------------------------------------------------------------- */
+ %exception {
+ try {
+ $action
+ } catch (PwrNlp::PwrNlpError &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ }
+ }
static TokenWriterPtr create_stream_writer(
const std::string& class_id_params,
std::ostream& os, const Tagset&
tagset);
+ %exception {
+ try {
+ $action
+ } catch (PwrNlp::PwrNlpError &e) {
+ PyErr_SetString(PyExc_IndexError, e.info().c_str());
+ return NULL;
+ }
+ }
static TokenWriterPtr create_path_writer(
const std::string& class_id_params,
const std::string& path,
diff --git a/swig/libpwrnlperror.i b/swig/libpwrnlperror.i
new file mode 100644
index 0000000000000000000000000000000000000000..31b715783ae28e70593b616c22dae756dc55e083
--- /dev/null
+++ b/swig/libpwrnlperror.i
@@ -0,0 +1,30 @@
+#ifndef SWIG_LIBPWRNLP_PWRNLPERROR_I
+#define SWIG_LIBPWRNLP_PWRNLPERROR_I
+
+%module libpwrnlperror
+%{
+ #include <libpwrutils/exception.h>
+%}
+
+%include "std_string.i"
+%include "std_except.i"
+
+%nodefaultctor PwrNlp::PwrNlpError;
+
+namespace PwrNlp {
+ class PwrNlpError : public std::runtime_error {
+ public:
+ PwrNlpError(const std::string &what);
+
+ /* --------------------------------------------------------------------- */
+ ~PwrNlpError() throw();
+
+ virtual std::string info() const;
+ virtual std::string scope() const;
+ };
+}
+
+using namespace std;
+using namespace PwrNlp;
+
+#endif /* SWIG_LIBPWRNLP_PWRNLPERROR_I */
diff --git a/swig/makewrapper.sh b/swig/makewrapper.sh
index fdf80c5f917ff9bff414341776ccf953e1e687d3..274daed608341da3aa6e8aa9a2326a2c94102a9b 100755
--- a/swig/makewrapper.sh
+++ b/swig/makewrapper.sh
@@ -2,7 +2,7 @@
if [ ${#} -eq 0 ]
then
- echo "Usage: $0 corpus2class"
+ echo -e "\e[1;31mUsage: $0 corpus2class\e[0m"
exit 1
fi