From a7afe1cf5e01dbfe402cbd7a7e35710013e69112 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Koco=C5=84?= <jan.kocon@pwr.edu.pl>
Date: Wed, 31 Oct 2018 16:43:09 +0100
Subject: [PATCH] added option to write Corpus2::Document to string (also in
SWIG)
---
libcorpus2/io/helpers.cpp | 36 ++++++++++++++++++++++++++++++++++++
libcorpus2/io/helpers.h | 15 +++++++++++++++
swig/tokenreader.i | 11 +++++++++++
3 files changed, 62 insertions(+)
diff --git a/libcorpus2/io/helpers.cpp b/libcorpus2/io/helpers.cpp
index 8ebdf85..c148d83 100644
--- a/libcorpus2/io/helpers.cpp
+++ b/libcorpus2/io/helpers.cpp
@@ -16,6 +16,8 @@ or FITNESS FOR A PARTICULAR PURPOSE.
#include <libcorpus2/io/helpers.h>
#include <libcorpus2/io/reader.h>
+#include <libcorpus2/io/writer.h>
+#include <boost/foreach.hpp>
#include <sstream>
namespace Corpus2 {
@@ -35,4 +37,38 @@ std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
return chunks;
}
+std::string write_chunk_to_utf8_string(
+ boost::shared_ptr<Chunk> c,
+ const Tagset& tagset,
+ const std::string& format)
+{
+ std::stringstream ss;
+ boost::shared_ptr<TokenWriter> writer = TokenWriter::create_stream_writer(
+ format, ss, tagset);
+ writer->write_chunk(*c);
+ writer->finish();
+ std::string data(std::istreambuf_iterator<char>(ss), {});
+ return data;
+}
+
+namespace whole {
+std::string write_document_to_utf8_string(
+ const Document& d,
+ const Tagset& tagset,
+ const std::string& format)
+{
+ std::stringstream ss;
+ boost::shared_ptr<TokenWriter> writer = TokenWriter::create_stream_writer(
+ format, ss, tagset);
+ BOOST_FOREACH(boost::shared_ptr<Chunk> c, d.paragraphs()) {
+ writer->write_chunk(*c);
+ }
+ writer->finish();
+ std::string data(std::istreambuf_iterator<char>(ss), {});
+ return data;
+}
+
+}
+
+
} /* end ns Corpus2 */
diff --git a/libcorpus2/io/helpers.h b/libcorpus2/io/helpers.h
index c61a904..34fe76f 100644
--- a/libcorpus2/io/helpers.h
+++ b/libcorpus2/io/helpers.h
@@ -19,6 +19,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
#include <libcorpus2/chunk.h>
#include <libcorpus2/tagset.h>
+#include <libcorpus2_whole/document.h>
namespace Corpus2 {
@@ -27,6 +28,20 @@ std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
const Tagset& tagset,
const std::string& format);
+std::string write_chunk_to_utf8_string(
+ boost::shared_ptr<Chunk> c,
+ const Tagset& tagset,
+ const std::string& format);
+
+namespace whole {
+
+std::string write_document_to_utf8_string(
+ const Document& d,
+ const Tagset& tagset,
+ const std::string& format);
+}
+
+
} /* end ns Corpus2 */
#endif // LIBCORPUS2_IO_HELPERS_H
diff --git a/swig/tokenreader.i b/swig/tokenreader.i
index 522c628..011f954 100644
--- a/swig/tokenreader.i
+++ b/swig/tokenreader.i
@@ -108,9 +108,20 @@ namespace Corpus2 {
std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
const std::string& data, const Tagset& tagset, const std::string& format);
+%feature("autodoc", "1");
+ std::string write_chunk_to_utf8_string(
+ boost::shared_ptr<Corpus2::Chunk> c, const Tagset& tagset, const std::string& format);
+namespace whole {
+%feature("autodoc", "1");
+ std::string write_document_to_utf8_string(
+ Corpus2::whole::Document& d, const Tagset& tagset, const std::string& format);
+}
+
+
}
using namespace std;
using namespace Corpus2;
+using namespace whole;
#endif /* SWIG_LIBCORPUS2_TOKENREADER_I */
--
GitLab