From e69b53e6d450809babf08c639fa0fdeb0c760a7e Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Fri, 18 Feb 2011 13:30:41 +0100
Subject: [PATCH] update AnnotationChannel interface, fix
make_iob_from_segments bug, upgrade make_annotation_vector, annotation heads
---
libcorpus2/ann/channel.cpp | 63 ++++++++++++++++++++++++++++++++------
libcorpus2/ann/channel.h | 31 ++++++++++++++++++-
2 files changed, 83 insertions(+), 11 deletions(-)
diff --git a/libcorpus2/ann/channel.cpp b/libcorpus2/ann/channel.cpp
index 5ff16f9..cb5fa31 100644
--- a/libcorpus2/ann/channel.cpp
+++ b/libcorpus2/ann/channel.cpp
@@ -1,16 +1,18 @@
#include <libcorpus2/ann/channel.h>
+#include <libpwrutils/foreach.h>
#include <algorithm>
#include <boost/bind.hpp>
+#include <sstream>
namespace Corpus2 {
AnnotationChannel::AnnotationChannel()
- : segments_(), iobs_()
+ : segments_(), iobs_(), heads_()
{
}
AnnotationChannel::AnnotationChannel(int size)
- : segments_(size), iobs_(size)
+ : segments_(size), iobs_(size), heads_(size)
{
}
@@ -20,7 +22,7 @@ void AnnotationChannel::make_iob_from_segments()
for (size_t i = 0; i < segments_.size(); ++i) {
if (segments_[i] == 0) {
iobs_[i] = IOB::O;
- } else if (segments_[i] != prev_seg) {
+ } else if (segments_[i] == prev_seg) {
iobs_[i] = IOB::I;
} else {
iobs_[i] = IOB::B;
@@ -67,6 +69,15 @@ int AnnotationChannel::renumber_segments()
return next;
}
+int AnnotationChannel::get_segment_at(int idx) const
+{
+ if (idx >= 0 && idx < static_cast<int>(segments_.size())) {
+ return segments_[idx];
+ } else {
+ return 0;
+ }
+}
+
IOB::Enum AnnotationChannel::get_iob_at(int idx)
{
if (idx >= 0 && idx < static_cast<int>(iobs_.size())) {
@@ -83,24 +94,56 @@ void AnnotationChannel::set_iob_at(int idx, IOB::Enum iob)
}
}
+bool AnnotationChannel::is_head_at(int idx) const
+{
+ if (idx >= 0 && idx < static_cast<int>(heads_.size())) {
+ return heads_[idx];
+ } else {
+ return false;
+ }
+}
+
+void AnnotationChannel::set_head_at(int idx, bool v)
+{
+ if (idx >= 0 && idx < static_cast<int>(heads_.size())) {
+ heads_[idx] = v;
+ }
+}
+
std::vector<Annotation> AnnotationChannel::make_annotation_vector() const
{
std::vector<Annotation> rv;
int smax = 0;
for (size_t i = 0; i < segments_.size(); ++i) {
int s = segments_[i];
- if (s > smax) {
- rv.resize(smax = s);
+ if (s > 0) {
+ if (s > smax) {
+ rv.resize(smax = s);
+ }
+ rv[s - 1].indices.push_back(i);
+ if (heads_[i]) {
+ rv[s - 1].head_index = i;
+ }
}
- rv[s].indices.push_back(i);
- if (rv[s].head_index == -1) {
- rv[s].head_index = i;
+ }
+ rv.erase(std::remove_if(rv.begin(), rv.end(),
+ boost::bind(&Annotation::empty, _1)), rv.end());
+ foreach (Annotation& a, rv) {
+ if (a.head_index == -1) {
+ a.head_index = a.indices[0];
}
}
std::sort(rv.begin(), rv.end(), AnnotationHeadCompare());
- rv.erase(std::remove_if(rv.begin(), rv.end(),
- boost::bind(&Annotation::empty, _1)));
return rv;
}
+std::string AnnotationChannel::dump_iob() const
+{
+ std::stringstream ss;
+ foreach (Corpus2::IOB::Enum e, iobs()) {
+ ss << Corpus2::IOB::to_string(e);
+ }
+ return ss.str();
+}
+
} /* end ns Corpus2 */
diff --git a/libcorpus2/ann/channel.h b/libcorpus2/ann/channel.h
index 0637638..cea905e 100644
--- a/libcorpus2/ann/channel.h
+++ b/libcorpus2/ann/channel.h
@@ -84,10 +84,22 @@ public:
/**
* The segment-index array accesor
*/
+ std::vector<int>& segments() {
+ return segments_;
+ }
+
+ /**
+ * The segment-index array accesor, const
+ */
const std::vector<int>& segments() const {
return segments_;
}
+ /**
+ * Segment index getter, 0 (no segment) if idx is out of range.
+ */
+ int get_segment_at(int idx) const;
+
/**
* The IOB data vector
*/
@@ -96,7 +108,7 @@ public:
}
/**
- * IOB getter, returns IOB::O if idx is out of range
+ * IOB getter, returns IOB::O if idx is out of range.
*/
IOB::Enum get_iob_at(int idx);
@@ -105,12 +117,29 @@ public:
*/
void set_iob_at(int idx, IOB::Enum iob);
+ /**
+ * Head flag getter, false if out of range.
+ */
+ bool is_head_at(int idx) const;
+
+ /**
+ * Head flag setter, out of range indices are not processed.
+ */
+ void set_head_at(int idx, bool v);
+
+ /**
+ * Compose a string consisting of all IOB markers in order.
+ */
+ std::string dump_iob() const;
+
private:
/// segment indices
std::vector<int> segments_;
/// IOB data
std::vector<IOB::Enum> iobs_;
+
+ std::vector<bool> heads_;
};
} /* end ns Corpus2 */
--
GitLab