From 844c9363bc44bd1fcb102d4f1f51f85ba571eb23 Mon Sep 17 00:00:00 2001
From: michal <michal@kalafior.(none)>
Date: Mon, 27 May 2013 10:12:01 +0200
Subject: [PATCH] fixed bug with reading documents without id - temporary
solution, whole DocumentReader class should be rebuilt
---
libcorpus2_whole/io/documentreader.cpp | 23 ++++++++++++++++++++++-
libcorpus2_whole/io/documentreader.h | 4 +++-
swig/documentreader.i | 1 +
3 files changed, 26 insertions(+), 2 deletions(-)
diff --git a/libcorpus2_whole/io/documentreader.cpp b/libcorpus2_whole/io/documentreader.cpp
index dcdf4f0..c65249d 100644
--- a/libcorpus2_whole/io/documentreader.cpp
+++ b/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;
diff --git a/libcorpus2_whole/io/documentreader.h b/libcorpus2_whole/io/documentreader.h
index 9dda867..b23a574 100644
--- a/libcorpus2_whole/io/documentreader.h
+++ b/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:
diff --git a/swig/documentreader.i b/swig/documentreader.i
index a360315..2290e46 100644
--- a/swig/documentreader.i
+++ b/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
--
GitLab