From 457811ce616780ff405356ed4652bbce616a896d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Pawe=C5=82=20K=C4=99dzia?= <Pawel.Kedzia@pwr.wroc.pl>
Date: Thu, 28 Apr 2011 09:26:32 +0200
Subject: [PATCH] Corpus2::Annotation wrapper
---
swig/Makefile | 25 ++++++---
swig/libcorpusannotationchannel.i | 84 +++++++++++++++++++++++++++++++
2 files changed, 103 insertions(+), 6 deletions(-)
create mode 100644 swig/libcorpusannotationchannel.i
diff --git a/swig/Makefile b/swig/Makefile
index 03b0031..04f2b15 100644
--- a/swig/Makefile
+++ b/swig/Makefile
@@ -21,7 +21,8 @@ CBIN=libcorpustag.o \
libcorpuschunk.o \
libcorpustokenwriter.o \
libcorpustokenreader.o \
- libcorpusiob.o
+ libcorpusiob.o \
+ libcorpusannotationchannel.o
CBINOUT=_boost_shared_ptr.so \
_libcorpustag.so \
@@ -33,7 +34,8 @@ CBINOUT=_boost_shared_ptr.so \
_libcorpuschunk.so \
_libcorpustokenwriter.so \
_libcorpustokenreader.so \
- _libcorpusiob.so
+ _libcorpusiob.so \
+ _libcorpusannotationchannel.so
CWRAP=boost_shared_ptr_wrap.cxx \
libcorpustag_wrap.cxx \
@@ -45,7 +47,8 @@ CWRAP=boost_shared_ptr_wrap.cxx \
libcorpuschunk_wrap.cxx \
libcorpustokenwriter_wrap.cxx \
libcorpustokenreader_wrap.cxx \
- libcorpusiob_wrap.cxx
+ libcorpusiob_wrap.cxx \
+ libcorpusannotationchannel_wrap.cxx
CWRAPBIN=boost_shared_ptr_wrap.o \
libcorpustag_wrap.o \
@@ -57,7 +60,8 @@ CWRAPBIN=boost_shared_ptr_wrap.o \
libcorpuschunk_wrap.o \
libcorpustokenwriter_wrap.o \
libcorpustokenreader_wrap.o \
- libcorpusiob_wrap.o
+ libcorpusiob_wrap.o \
+ libcorpusannotationchannel_wrap.o
PYMODULES=boost_shared_ptr.py \
libcorpustag.py \
@@ -69,7 +73,8 @@ PYMODULES=boost_shared_ptr.py \
libcorpuschunk.py \
libcorpustokenwriter.py \
libcorpustokenreader.py \
- libcorpusiob.py
+ libcorpusiob.py \
+ libcorpusannotationchannel.py
PYCBIN=boost_shared_ptr.pyc \
libcorpustag.pyc \
@@ -81,7 +86,8 @@ PYCBIN=boost_shared_ptr.pyc \
libcorpuschunk.pyc \
libcorpustokenwriter.pyc \
libcorpustokenreader.pyc \
- libcorpusiob.pyc
+ libcorpusiob.pyc \
+ libcorpusannotationchannel.pyc
# -----------------------------------------------------------------------------
all:boost_shared_ptr.o $(CBIN)
@@ -164,6 +170,13 @@ libcorpusiob.o:
$(CPP) -shared libcorpusiob_wrap.o \
$(PWRUTILBIN) $(CORPUS2BIN) -o _libcorpusiob.so
+# AnnotationChannel
+libcorpusannotationchannel.o:
+ $(SWIG) $(SWIGOPTS_LANG) libcorpusannotationchannel.i
+ $(CPP) -c libcorpusannotationchannel_wrap.cxx -I$(PYTHONDIR) $(CPPFLAGS)
+ $(CPP) -shared libcorpusannotationchannel_wrap.o \
+ $(PWRUTILBIN) $(CORPUS2BIN) -o _libcorpusannotationchannel.so
+
# -----------------------------------------------------------------------------
clean:
rm -f $(CBIN) $(CBINOUT) $(CWRAP) $(CWRAPBIN) $(PYMODULES) $(PYCBIN)
diff --git a/swig/libcorpusannotationchannel.i b/swig/libcorpusannotationchannel.i
new file mode 100644
index 0000000..91a7929
--- /dev/null
+++ b/swig/libcorpusannotationchannel.i
@@ -0,0 +1,84 @@
+#ifndef SWIG_LIBCORPUS2_ANNOTATIONCHANNEL_I
+#define SWIG_LIBCORPUS2_ANNOTATIONCHANNEL_I
+
+%module libcorpusannotationchannel
+%{
+ #include <libcorpus2/ann/channel.h>
+%}
+
+%include "std_string.i"
+%include "std_vector.i"
+%include "libcorpusiob.i"
+
+%template(IntVector) std::vector<int>;
+%template(IOBVector) std::vector<Corpus2::IOB::Enum>;
+%template(AnnotationVector) std::vector<Corpus2::Annotation>;
+
+namespace Corpus2 {
+ struct Annotation {
+ Annotation() : indices(), head_index(-1) {}
+ std::vector<int> indices;
+ int head_index;
+ bool empty() const;
+ bool sane() const;
+ };
+
+ class AnnotationChannel {
+ public:
+ AnnotationChannel();
+ explicit AnnotationChannel(int size);
+ ~AnnotationChannel();
+
+ /* --------------------------------------------------------------------- */
+
+ int size() const;
+ void resize(int size);
+
+ void make_iob_from_segments();
+ void make_segments_from_iob();
+
+ int renumber_segments();
+ int get_new_segment_index() const;
+
+ /* --------------------------------------------------------------------- */
+ enum AnnotationVectorMode {
+ O_DISJOINT_EXCLUSIVE = 0,
+ O_CONTINUOUS = 1,
+ O_INCLUSIVE = 2,
+ O_CONTINUOUS_INCLUSIVE = 3
+ };
+
+ std::vector<Annotation> make_annotation_vector(
+ AnnotationVectorMode mode = O_DISJOINT_EXCLUSIVE) const;
+
+ /* --------------------------------------------------------------------- */
+ std::vector<int>& segments();
+ const std::vector<int>& segments() const;
+
+ int get_segment_at(int idx) const;
+ void set_segment_at(int token_idx, int segment_idx);
+
+ /* --------------------------------------------------------------------- */
+ const std::vector<IOB::Enum>& iobs() const;
+ IOB::Enum get_iob_at(int idx);
+
+ /* --------------------------------------------------------------------- */
+ void set_iob_at(int idx, IOB::Enum iob);
+ bool is_head_at(int idx) const;
+ void set_head_at(int idx, bool v);
+
+ /* --------------------------------------------------------------------- */
+ std::string dump_iob() const;
+ std::string dump_segments() const;
+ std::string dump_heads() const;
+ std::string dump_alpha() const;
+
+ /* --------------------------------------------------------------------- */
+ void do_counts(int& annotations, int& disjoint, int& unannotated) const;
+ };
+}
+
+using namespace std;
+using namespace Corpus2;
+
+#endif /* SWIG_LIBCORPUS2_ANNOTATIONCHANNEL_I */
--
GitLab