corpus2 helper to read a string

libcorpus2/CMakeLists.txt

@@ -57,6 +57,7 @@ SET(libcorpus2_STAT_SRC
 	tokenmetadata.cpp
 	io/cclreader.cpp
 	io/cclwriter.cpp
+	io/helpers.cpp
 	io/fastxces.cpp
 	io/nonewriter.cpp
 	io/orthwriter.cpp

libcorpus2/io/helpers.cpp

+#include <libcorpus2/io/helpers.h>
+#include <libcorpus2/io/reader.h>
+#include <sstream>
+namespace Corpus2 {
+
+std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
+        const std::string& data,
+        const Tagset& tagset,
+        const std::string& format)
+{
+	std::stringstream ss;
+	ss << data;
+	boost::shared_ptr<TokenReader> reader = TokenReader::create_stream_reader(
+		format, tagset, ss);
+	std::vector<boost::shared_ptr<Chunk> > chunks;
+	while (boost::shared_ptr<Chunk> c = reader->get_next_chunk()) {
+		chunks.push_back(c);
+	}
+	return chunks;
+}
+
+} /* end ns Corpus2 */

libcorpus2/io/helpers.h

+#ifndef LIBSORPUS2_IO_HELPERS_H
+#define LIBCORPUS2_IO_HELPERS_H
+
+#include <libcorpus2/chunk.h>
+#include <libcorpus2/tagset.h>
+
+namespace Corpus2 {
+
+std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
+	const std::string& data,
+	const Tagset& tagset,
+	const std::string& format);
+
+} /* end ns Corpus2 */
+
+#endif // LIBCORPUS2_IO_HELPERS_H

swig/libcorpuschunk.i

@@ -46,6 +46,7 @@ namespace Corpus2 {
   };
 }
 
+%template(ChunkPtrVector) std::vector<boost::shared_ptr<Chunk> >;
 using namespace std;
 using namespace Corpus2;
 

swig/libcorpustokenreader.i

@@ -4,6 +4,7 @@
 %module libcorpustokenreader
 %{
   #include <libcorpus2/io/reader.h>
+  #include <libcorpus2/io/helpers.h>
 %}
 
 %include "libcorpustag.i"
@@ -73,6 +74,10 @@ namespace Corpus2 {
     static std::string reader_help(const std::string& class_id);
     static std::vector<std::string> available_reader_types_help();
   };
+
+  std::vector<boost::shared_ptr<Chunk> > read_chunks_from_utf8_string(
+    const std::string& data, const Tagset& tagset, const std::string& format);
+ 
 }
 
 using namespace std;