From 2702b4e1dc855ab2cce727caabf599245dd9919c Mon Sep 17 00:00:00 2001
From: ilor <kailoran@gmail.com>
Date: Mon, 15 Nov 2010 10:43:31 +0100
Subject: [PATCH] Implement TSet:: is_subset_of, equals and intersects Change
StrSet::is_subset_of to correctly handle empty sets, add some comments Add
Tset::insert_symbol Add tests for new functions Bump required corpus2 version
to 0.1.2 due to a fixed bug there
---
libwccl/CMakeLists.txt | 2 +-
libwccl/values/strset.cpp | 19 +++++++------
libwccl/values/strset.h | 8 ++++++
libwccl/values/tset.cpp | 5 ++++
libwccl/values/tset.h | 25 ++++++++++------
tests/values.cpp | 60 +++++++++++++++++++++++++++++++++++++++
6 files changed, 101 insertions(+), 18 deletions(-)
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index b69bbb4..8c68a29 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -4,7 +4,7 @@ PROJECT(wccl)
include_directories( ${CMAKE_CURRENT_BINARY_DIR}/include/ )
-find_package(Corpus2 0.1.1 REQUIRED)
+find_package(Corpus2 0.1.2 REQUIRED)
set(LIBS ${LIBS} ${Corpus2_LIBRARY})
find_package(PwrUtils 0.0.3 REQUIRED)
diff --git a/libwccl/values/strset.cpp b/libwccl/values/strset.cpp
index ed6b4bc..02489f4 100644
--- a/libwccl/values/strset.cpp
+++ b/libwccl/values/strset.cpp
@@ -24,29 +24,32 @@ std::string StrSet::to_raw_string() const
}
bool StrSet::intersects(const StrSet &other) const {
- if(empty() || other.empty()) {
+ if (empty() || other.empty()) {
return false;
}
- //we just want to check if there is an intersection, no
+ //We just want to check if there is an intersection, no
//need to actually compute it to check if it's empty.
- //doing it like below sounds faster than, say, sorting
- //the sets and using set_intersection
+ //Doing it like below sounds faster than, say, sorting
+ //the sets and using set_intersection.
+ //it's faster to iterate through the smaller set and check in
+ //the larger than it is to do the opposite, hence the &?: below
const set_t& smaller = size() < other.size() ? set_ : other.set_;
const set_t& bigger = size() < other.size() ? other.set_ : set_;
foreach (const UnicodeString& u, smaller) {
- if(bigger.find(u) != bigger.end()) {
+ if (bigger.find(u) != bigger.end()) {
return true;
}
}
return false;
}
-bool StrSet::is_subset_of(const StrSet &other) const {
- if(empty() || size() > other.size()) {
+bool StrSet::is_subset_of(const StrSet &other) const
+{
+ if (size() > other.size()) {
return false;
}
foreach (const UnicodeString& u, set_) {
- if(other.set_.find(u) == other.set_.end()) {
+ if (other.set_.find(u) == other.set_.end()) {
return false;
}
}
diff --git a/libwccl/values/strset.h b/libwccl/values/strset.h
index e984bd0..3545e80 100644
--- a/libwccl/values/strset.h
+++ b/libwccl/values/strset.h
@@ -56,8 +56,16 @@ public:
return set_.empty();
}
+ /**
+ * @return true if each string from this set exists in the other set
+ * (note that an empty set is a subset of anything)
+ */
bool is_subset_of(const StrSet& other) const;
+ /**
+ * @return true if there is at least one common string between this set and
+ * the other set (an empty set intersects with nothing)
+ */
bool intersects(const StrSet& other) const;
bool equals(const StrSet& other) const {
diff --git a/libwccl/values/tset.cpp b/libwccl/values/tset.cpp
index 9573bcb..8b1090e 100644
--- a/libwccl/values/tset.cpp
+++ b/libwccl/values/tset.cpp
@@ -15,4 +15,9 @@ std::string TSet::to_string(const Corpus2::Tagset& tagset) const
return tagset.tag_to_symbol_string(tag_);
}
+void TSet::insert_symbol(const Corpus2::Tagset& tagset, const std::string& s)
+{
+ tag_.combine_with(tagset.parse_symbol(s));
+}
+
} /* end ns Wccl */
diff --git a/libwccl/values/tset.h b/libwccl/values/tset.h
index 3c7d125..6d6813b 100644
--- a/libwccl/values/tset.h
+++ b/libwccl/values/tset.h
@@ -33,23 +33,30 @@ public:
return tag_;
}
+ void insert_symbol(const Corpus2::Tagset& tagset, const std::string& s);
+
bool empty() const {
return tag_.is_null();
}
- bool is_subset_of(const TSet& /*tset*/) const {
- //TODO: implement this
- return false;
+ /**
+ * @return true if each tagset symbol from this set exists in the other set
+ * (note that an empty set is a subset of anything)
+ */
+ bool is_subset_of(const TSet& other) const {
+ return tag_.get_masked(other.tag_) == tag_;
}
- bool intersects(const TSet& /*tset*/) const {
- //TODO: implement this
- return false;
+ /**
+ * @return true if there is at least one common symbol between this set and
+ * the other set (an empty set intersects with nothing)
+ */
+ bool intersects(const TSet& other) const {
+ return !tag_.get_masked(other.tag_).is_null();
}
- bool equals(const TSet& /*tset*/) const {
- //TODO: implement this
- return false;
+ bool equals(const TSet& other) const {
+ return tag_ == other.tag_;
}
std::string to_string(const Corpus2::Tagset &) const;
diff --git a/tests/values.cpp b/tests/values.cpp
index 6d2ac2d..78df7ec 100644
--- a/tests/values.cpp
+++ b/tests/values.cpp
@@ -1,5 +1,6 @@
#include <boost/test/unit_test.hpp>
#include <boost/bind.hpp>
+#include <libcorpus2/tagsetmanager.h>
#include <libwccl/variables.h>
@@ -25,4 +26,63 @@ BOOST_AUTO_TEST_CASE(tsetz)
BOOST_CHECK_EQUAL(v.get_type_name(), TSet::type_name);
}
+BOOST_AUTO_TEST_CASE(strset_ops)
+{
+ StrSet s1, s2;
+ BOOST_CHECK(s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s1.insert_utf8("aaa");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(!s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s2.insert_utf8("bbb");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(!s1.is_subset_of(s2));
+ BOOST_CHECK(!s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s2.insert_utf8("aaa");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(!s2.is_subset_of(s1));
+ BOOST_CHECK(s1.intersects(s2));
+ s1.insert_utf8("bbb");
+ BOOST_CHECK(s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(s1.intersects(s2));
+}
+
+BOOST_AUTO_TEST_CASE(tset_ops)
+{
+ TSet s1, s2;
+ const Corpus2::Tagset& tagset = Corpus2::get_named_tagset("kipi");
+ BOOST_CHECK(s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s1.insert_symbol(tagset, "subst");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(!s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s2.insert_symbol(tagset, "pl");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(!s1.is_subset_of(s2));
+ BOOST_CHECK(!s2.is_subset_of(s1));
+ BOOST_CHECK(!s1.intersects(s2));
+ s2.insert_symbol(tagset, "subst");
+ BOOST_CHECK(!s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(!s2.is_subset_of(s1));
+ BOOST_CHECK(s1.intersects(s2));
+ s1.insert_symbol(tagset, "pl");
+ BOOST_CHECK(s1.equals(s2));
+ BOOST_CHECK(s1.is_subset_of(s2));
+ BOOST_CHECK(s2.is_subset_of(s1));
+ BOOST_CHECK(s1.intersects(s2));
+}
+
BOOST_AUTO_TEST_SUITE_END()
--
GitLab