From 75b314aa92c378b4b65b85726596bfa2899dfa79 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Pawe=C5=82=20K=C4=99dzia?= <Pawel.Kedzia@pwr.wroc.pl>
Date: Tue, 3 Jan 2012 12:36:44 +0100
Subject: [PATCH] Document contains path to file what was readed.
---
libcorpus2_whole/document.cpp | 4 ++--
libcorpus2_whole/document.h | 17 ++++++++++++++---
libcorpus2_whole/io/documentreader.cpp | 9 ++++++++-
libcorpus2_whole/io/documentreader.h | 10 ++++++++++
swig/document.i | 4 +++-
5 files changed, 37 insertions(+), 7 deletions(-)
diff --git a/libcorpus2_whole/document.cpp b/libcorpus2_whole/document.cpp
index 6b949b3..9b391b8 100644
--- a/libcorpus2_whole/document.cpp
+++ b/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)
{
}
diff --git a/libcorpus2_whole/document.h b/libcorpus2_whole/document.h
index 97ad4b8..d02a043 100644
--- a/libcorpus2_whole/document.h
+++ b/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
diff --git a/libcorpus2_whole/io/documentreader.cpp b/libcorpus2_whole/io/documentreader.cpp
index 4d35ffe..8aa1a79 100644
--- a/libcorpus2_whole/io/documentreader.cpp
+++ b/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) {
diff --git a/libcorpus2_whole/io/documentreader.h b/libcorpus2_whole/io/documentreader.h
index d0c4af3..8b092de 100644
--- a/libcorpus2_whole/io/documentreader.h
+++ b/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
diff --git a/swig/document.i b/swig/document.i
index 900053a..0e8ea3c 100644
--- a/swig/document.i
+++ b/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
--
GitLab