Document contains path to file what was readed.

libcorpus2_whole/document.cpp

@@ -20,8 +20,8 @@ or FITNESS FOR A PARTICULAR PURPOSE.
 namespace Corpus2 {
 namespace whole {
 
-Document::Document()
-	: paragraphs_(), relations_()
+Document::Document(const std::string& path)
+	: paragraphs_(), relations_(), path_(path)
 {
 }
 

libcorpus2_whole/document.h

@@ -33,7 +33,10 @@ namespace whole {
 class Document
 {
 public:
-	Document();
+	/**
+	 * Path to file, if not set, then default is empty
+	 */
+	Document(const std::string& path = "");
 	~Document();
 
 	/// Adds paragraphs to document
@@ -56,12 +59,20 @@ public:
 		return relations_;
 	}
 
+	/// Returns path to the document
+	const std::string& path() const {
+		return path_;
+	}
+
 protected:
 	/// Paragraphs in document
-	std::vector< boost::shared_ptr<Chunk> > paragraphs_;
+	std::vector<boost::shared_ptr<Chunk> > paragraphs_;
 
 	/// Relations in document
-	std::vector< boost::shared_ptr<Relation> > relations_;
+	std::vector<boost::shared_ptr<Relation> > relations_;
+
+	/// Path to the file (if it's not a file, then is empty)
+	const std::string path_;
 };
 
 } // whole ns

libcorpus2_whole/io/documentreader.cpp

@@ -24,6 +24,7 @@ namespace whole {
 			: DocumentReaderI("document")
 	{
 		make_readers(tagset, annot_path, rela_path);
+		make_id_doc(annot_path, rela_path);
 	}
 
 	void DocumentReader::make_readers(const Tagset& tagset,
@@ -33,10 +34,16 @@ namespace whole {
 		rel_reader_ = boost::make_shared<RelationReader>(rela_path);
 	}
 
+	void DocumentReader::make_id_doc(const std::string &annot_path,
+		const std::string &rela_path)
+	{
+		id_ = (annot_path + ";" + rela_path);
+	}
+
 	boost::shared_ptr<Document> DocumentReader::read()
 	{
 		boost::shared_ptr<Chunk> chunk;
-		boost::shared_ptr<Document> document = boost::make_shared<Document>();
+		boost::shared_ptr<Document> document = boost::make_shared<Document>(id_);
 
 		// Read ccl document and makes document
 		while (1) {

libcorpus2_whole/io/documentreader.h

@@ -78,12 +78,22 @@ private:
 			const std::string &annot_path,
 			const std::string &rela_path);
 
+	/**
+	 * Based on given paths (annotations and relations) makes document identifier
+	 * Document identifier is set to id_ class-state
+	 */
+	void make_id_doc(const std::string &annot_path,
+					 const std::string &rela_path);
+
 	// -------------------------------------------------------------------------
 	/// Pointer to CclReader
 	boost::shared_ptr<CclReader> ccl_reader_;
 
 	/// Pointer to RelationReader
 	boost::shared_ptr<RelationReader> rel_reader_;
+
+	/// Future document identifier
+	std::string id_;
 };
 
 } // whole ns

swig/document.i

@@ -19,7 +19,7 @@ namespace Corpus2 {
 namespace whole {
   class Document {
   public:
-    Document();
+    Document(const std::string& path = "");
     ~Document();
 
     void add_paragraph(const boost::shared_ptr<Chunk> para);
@@ -27,6 +27,8 @@ namespace whole {
 
     const std::vector< boost::shared_ptr<Chunk> >& paragraphs() const;
     const std::vector< boost::shared_ptr<Relation> >& relations() const;
+
+    const std::string& path() const;
   };
 } // whole ns
 } // Corpus2 ns