From b6b6351e18173d19bef591a6f893e6a5e3e1a254 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 13:26:29 +0100
Subject: [PATCH] Sentence id handling in cclreader/writer
---
libcorpus2/io/cclreader.cpp | 13 +++++++++++--
libcorpus2/io/cclwriter.cpp | 10 +++++++++-
2 files changed, 20 insertions(+), 3 deletions(-)
diff --git a/libcorpus2/io/cclreader.cpp b/libcorpus2/io/cclreader.cpp
index 49fcc03..f796eb2 100644
--- a/libcorpus2/io/cclreader.cpp
+++ b/libcorpus2/io/cclreader.cpp
@@ -140,9 +140,18 @@ void CclReaderImpl::start_chunk(const AttributeList& attributes)
-void CclReaderImpl::start_sentence(const AttributeList& /*attributes*/)
+void CclReaderImpl::start_sentence(const AttributeList &attributes)
{
- ann_sent_ = boost::make_shared<AnnotatedSentence>();
+ // find sentence id
+ std::string id = "";
+ foreach (const Attribute& a, attributes) {
+ if (a.name == "id") {
+ id = a.value;
+ break;
+ }
+ }
+
+ ann_sent_ = boost::make_shared<AnnotatedSentence>(id);
sent_ = ann_sent_;
state_ = STATE_SENTENCE;
}
diff --git a/libcorpus2/io/cclwriter.cpp b/libcorpus2/io/cclwriter.cpp
index 221d6a3..4909291 100644
--- a/libcorpus2/io/cclwriter.cpp
+++ b/libcorpus2/io/cclwriter.cpp
@@ -33,7 +33,15 @@ void CclWriter::write_sentence(const Sentence& s)
void CclWriter::write_sentence_int(const Sentence &s)
{
const AnnotatedSentence* ann = dynamic_cast<const AnnotatedSentence*>(&s);
- osi() << "<sentence>\n";
+
+ std::string id = s.id();
+ if (id == "") {
+ osi() << "<sentence>\n";
+ }
+ else {
+ osi() << "<sentence id=\"" << id << "\">\n";
+ }
+
if (use_indent_) indent_more();
for (size_t idx = 0; idx < s.size(); ++idx) {
const Token* t = s.tokens()[idx];
--
GitLab