From 12ed6c78436783a45b797b576406680f72a75a7d Mon Sep 17 00:00:00 2001
From: Adam Radziszewski <adam.radziszewski@pwr.wroc.pl>
Date: Thu, 17 Nov 2011 13:28:55 +0100
Subject: [PATCH] make Annotation aware of its original segment idx
---
libcorpus2/ann/channel.cpp | 1 +
libcorpus2/ann/channel.h | 6 +++++-
swig/annotationchannel.i | 3 ++-
tests/ann_basic.cpp | 4 ++++
4 files changed, 12 insertions(+), 2 deletions(-)
diff --git a/libcorpus2/ann/channel.cpp b/libcorpus2/ann/channel.cpp
index 34e7549..581c9dd 100644
--- a/libcorpus2/ann/channel.cpp
+++ b/libcorpus2/ann/channel.cpp
@@ -152,6 +152,7 @@ std::vector<Annotation> AnnotationChannel::make_annotation_vector(
rv.resize(smax = s);
}
rv[s - 1].indices.push_back(i);
+ rv[s - 1].seg_number = s;
if (heads_[i]) {
rv[s - 1].head_index = i;
}
diff --git a/libcorpus2/ann/channel.h b/libcorpus2/ann/channel.h
index 204ee2c..d0ba721 100644
--- a/libcorpus2/ann/channel.h
+++ b/libcorpus2/ann/channel.h
@@ -13,12 +13,16 @@ namespace Corpus2 {
/**
* A general indexed annotation spanning a possibly disjoint group of
* tokens, with a distinguished 'head' token.
+ * The annotation contains information about token indices (array subscripts
+ * referring to the original sentence), the index of a token marked as head,
+ * as well as the original segment number (segment index).
*/
struct Annotation
{
- Annotation() : indices(), head_index(-1) {}
+ Annotation() : indices(), head_index(-1), seg_number(0) {}
std::vector<int> indices;
int head_index;
+ int seg_number;
bool empty() const {
return indices.empty();
}
diff --git a/swig/annotationchannel.i b/swig/annotationchannel.i
index 1d9c8da..8ffeb1d 100644
--- a/swig/annotationchannel.i
+++ b/swig/annotationchannel.i
@@ -18,9 +18,10 @@
namespace Corpus2 {
struct Annotation {
- Annotation() : indices(), head_index(-1) {}
+ Annotation() : indices(), head_index(-1), seg_number(0) {}
std::vector<int> indices;
int head_index;
+ int seg_number;
bool empty() const;
bool sane() const;
};
diff --git a/tests/ann_basic.cpp b/tests/ann_basic.cpp
index 54a1e0d..5a4204b 100644
--- a/tests/ann_basic.cpp
+++ b/tests/ann_basic.cpp
@@ -119,6 +119,10 @@ BOOST_AUTO_TEST_CASE( make_ann )
BOOST_CHECK_EQUAL(anns[2].indices[1], 4);
BOOST_CHECK_EQUAL(anns[3].indices[0], 5);
BOOST_CHECK_EQUAL(anns[2].head_index, 3);
+ BOOST_CHECK_EQUAL(anns[0].seg_number, 1);
+ BOOST_CHECK_EQUAL(anns[1].seg_number, 2);
+ BOOST_CHECK_EQUAL(anns[2].seg_number, 3);
+ BOOST_CHECK_EQUAL(anns[3].seg_number, 4);
}
--
GitLab