Merge branch 'master' of nlp.pwr.wroc.pl:corpus2

libcorpus2_whole/io/documentreader.cpp

@@ -73,8 +73,29 @@ boost::shared_ptr<Document> DocumentReader::read()
 	throw Corpus2Error(corpus_type_ + " is an unknown reader type!");
 }
 
+boost::shared_ptr<Document> DocumentReader::read_with_auto_id()
+{
+	std::string line;
+#ifdef WITH_POLIQARP
+	if (corpus_type_ == "poliqarp") {
+		return this->reader->read();
+	}
+#endif
+	if (corpus_type_ == "document") {
+		if (std::getline(corpus_file, line)) {
+			boost::shared_ptr<CclRelReader> cclrel_reader = get_cclrel_reader(line);
+			cclrel_reader->get_option("autogen_sent_id");
+			cclrel_reader->set_option("autogen_chunk_id");
+			return cclrel_reader->read();
+		} else {
+			return boost::make_shared<Document>("End");
+		}
+	}
+	throw Corpus2Error(corpus_type_ + " is an unknown reader type!");
+}
+
 
-boost::shared_ptr<DocumentReaderI> DocumentReader::get_cclrel_reader(std::string& line)
+boost::shared_ptr<CclRelReader> DocumentReader::get_cclrel_reader(std::string& line)
 {
 	std::string ann_path, rel_path;
 

libcorpus2_whole/io/documentreader.h

@@ -20,6 +20,7 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 #include <string>
 #include <libcorpus2_whole/corpus.h>
 #include <libcorpus2_whole/io/reader_i.h>
+#include <libcorpus2_whole/io/cclrelreader.h>
 
 namespace Corpus2 {
 namespace whole {
@@ -30,10 +31,11 @@ public:
 	DocumentReader(const Tagset& tagset, const std::string& corpus_type, const std::string& corpus_file_path, const std::string& corpus_reader);
 	DocumentReader(const Tagset& tagset, const std::string& corpus_type, const std::string& corpus_file_path);
 	boost::shared_ptr<Document> read();
+	boost::shared_ptr<Document> read_with_auto_id();
 
 private:
 
-	boost::shared_ptr<DocumentReaderI> get_cclrel_reader(std::string& line);
+	boost::shared_ptr<CclRelReader> get_cclrel_reader(std::string& line);
 	boost::shared_ptr<DocumentReaderI> reader;
         std::ifstream corpus_file;
 private:

swig/documentreader.i

@@ -22,6 +22,7 @@ namespace whole {
     DocumentReader(const Tagset& tagset, const std::string& corpus_type,const std::string& corpus_file_path, const std::string& corpus_reader);
     DocumentReader(const Tagset& tagset, const std::string& corpus_type,const std::string& corpus_file_path);
     boost::shared_ptr<Document> read();
+    boost::shared_ptr<Document> read_with_auto_id();
   };
 } // whole ns
 } // Corpus2 ns

swig/token.i

@@ -67,6 +67,8 @@ namespace Corpus2 {
     boost::shared_ptr<TokenMetaData> get_metadata() const;
     void set_metadata(TokenMetaData& md);
     void set_metadata_ptr(boost::shared_ptr<TokenMetaData> md);
+
+    void create_metadata();
   };
 
   %extend Token {