From 7bb4fa18b05368e490c5639871b99a1603d48cc9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Pawe=C5=82=20K=C4=99dzia?= <Pawel.Kedzia@pwr.wroc.pl>
Date: Tue, 15 Nov 2011 15:59:27 +0100
Subject: [PATCH] DocumentReader implementation
---
libcorpus2/io/docreader.cpp | 33 ++++++++++++++++++++++++++++-
libcorpus2/io/docreader.h | 42 ++++++++++++++++++++++++++++++++-----
2 files changed, 69 insertions(+), 6 deletions(-)
diff --git a/libcorpus2/io/docreader.cpp b/libcorpus2/io/docreader.cpp
index 597190c..463ae4f 100644
--- a/libcorpus2/io/docreader.cpp
+++ b/libcorpus2/io/docreader.cpp
@@ -14,9 +14,40 @@ or FITNESS FOR A PARTICULAR PURPOSE.
See the LICENSE and COPYING files for more details.
*/
-#include <libcorpus2/io/docreader.h>
#include <boost/make_shared.hpp>
+#include <libcorpus2/io/docreader.h>
namespace Corpus2 {
+ DocumentReader::DocumentReader(const Tagset& tagset,
+ const std::string &annot_path, const std::string &rela_path)
+ {
+ make_readers(tagset, annot_path, rela_path);
+ }
+
+ void DocumentReader::make_readers(const Tagset& tagset,
+ const std::string &annot_path, const std::string &rela_path)
+ {
+ ccl_reader_ = boost::make_shared<CclReader>(tagset, annot_path);
+ rel_reader_ = boost::make_shared<RelationReader>(rela_path);
+ }
+
+ boost::shared_ptr<Document> DocumentReader::read()
+ {
+ boost::shared_ptr<Chunk> chunk;
+ boost::shared_ptr<Document> document = boost::make_shared<Document>();
+
+ // Read ccl document and makes document
+ while (1) {
+ chunk = ccl_reader_->get_next_chunk();
+ if (!chunk) {
+ break;
+ }
+ else {
+ document->add_paragraph(chunk);
+ }
+ }
+
+ return document;
+ }
} /* end ns Corpus2 */
diff --git a/libcorpus2/io/docreader.h b/libcorpus2/io/docreader.h
index 632def6..b769c56 100644
--- a/libcorpus2/io/docreader.h
+++ b/libcorpus2/io/docreader.h
@@ -17,7 +17,11 @@ or FITNESS FOR A PARTICULAR PURPOSE.
#ifndef LIBCORPUS2_DOCREADER_H
#define LIBCORPUS2_DOCREADER_H
-#include <libcorpus2/io/reader.h>
+#include <libcorpus2/document.h>
+#include <libcorpus2/io/cclreader.h>
+#include <libcorpus2/io/relreader.h>
+
+#include <boost/shared_ptr.hpp>
namespace Corpus2 {
@@ -32,11 +36,39 @@ public:
* chunk-style annotations are read from annot_path, while relations
* between chunk-style annotations are read from rela_path.
* Both path may in particular point to the same path.
- * TODO!
+ * @param tagset Tagset to use
+ * @param annot_path Path to file with morphosyntax and chunk-style annotations
+ * @param rela_path path to file with relations
*/
- DocumentReader(const std::string &annot_path,
- const std::string &rela_path,
- const std::string &rdr_class_id = "ccl");
+ DocumentReader(const Tagset& tagset,
+ const std::string &annot_path,
+ const std::string &rela_path);
+
+ /**
+ * Reads document stored in given file(s), in file with morphosyntax and
+ * chunk-style annotations and from file with relations.
+ * @return Pointer to readed Document
+ */
+ boost::shared_ptr<Document> read();
+
+private:
+ /**
+ * Makes CclReader and RelationReader for given paths to files.
+ * @param annot_path Path to file with morphosyntax and chunk-style annotations
+ * @param tagset Tagset to use in CclReader
+ * @param rela_path path to file with relations
+ */
+ void make_readers(
+ const Tagset& tagset,
+ 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_;
};
} /* end ns Corpus2 */
--
GitLab