From ec4caf8d6d63444f90a277d388d0f92b48891473 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Adam=20Wardy=C5=84ski?= <no@email>
Date: Wed, 10 Nov 2010 21:02:10 +0100
Subject: [PATCH] Adding "upper" operator as ToUpper class
---
libwccl/CMakeLists.txt | 1 +
libwccl/ops/toupper.cpp | 26 ++++++++++++++++++
libwccl/ops/toupper.h | 56 +++++++++++++++++++++++++++++++++++++++
tests/strsetfunctions.cpp | 53 ++++++++++++++++++++++++++++++++++++
4 files changed, 136 insertions(+)
create mode 100644 libwccl/ops/toupper.cpp
create mode 100644 libwccl/ops/toupper.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index b142613..c5fb97c 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -22,6 +22,7 @@ SET(libwccl_STAT_SRC
ops/or.cpp
ops/predicate.cpp
ops/tolower.cpp
+ ops/toupper.cpp
parser/Parser.cpp
parser/ParserException.cpp
sentencecontext.cpp
diff --git a/libwccl/ops/toupper.cpp b/libwccl/ops/toupper.cpp
new file mode 100644
index 0000000..1e94d31
--- /dev/null
+++ b/libwccl/ops/toupper.cpp
@@ -0,0 +1,26 @@
+#include <libwccl/ops/toupper.h>
+#include <libwccl/ops/formatters.h>
+
+namespace Wccl {
+
+std::string ToUpper::to_string(const Corpus2::Tagset& tagset) const
+{
+ return UnaryFunctionFormatter::to_string(tagset, *this, *strset_expr_);
+}
+
+std::string ToUpper::to_raw_string() const {
+ return UnaryFunctionFormatter::to_raw_string(*this, *strset_expr_);
+}
+
+ToUpper::BaseRetValPtr ToUpper::apply_internal(const SentenceContext& context) const {
+ const boost::shared_ptr<StrSet >& set = strset_expr_->apply(context);
+ boost::shared_ptr<StrSet > u_set = boost::make_shared<StrSet>();
+ //TODO: should tolower be a method of StrSet as well?
+ foreach(const UnicodeString& s, set->contents()) {
+ //TODO: what about locale? is default ok? should the context hold it?
+ u_set->insert(UnicodeString(s).toUpper());
+ }
+ return u_set;
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/toupper.h b/libwccl/ops/toupper.h
new file mode 100644
index 0000000..6491d92
--- /dev/null
+++ b/libwccl/ops/toupper.h
@@ -0,0 +1,56 @@
+#ifndef LIBWCCL_OPS_TOUPPER_H
+#define LIBWCCL_OPS_TOUPPER_H
+
+#include <boost/shared_ptr.hpp>
+#include <libwccl/values/strset.h>
+#include <libwccl/ops/functions.h>
+
+namespace Wccl {
+
+/**
+ * Operator that takes a set of strings and returns a new
+ * set with corresponding values in upper case form
+ */
+class ToUpper : public Function<StrSet> {
+public:
+ typedef boost::shared_ptr<Function<StrSet> > StrSetFunctionPtr;
+
+ ToUpper(const StrSetFunctionPtr& strset_expr)
+ : strset_expr_(strset_expr)
+ {
+ BOOST_ASSERT(strset_expr_);
+ }
+
+ /**
+ * String representation of the operator in form of:
+ * "upper(strset_expr_string)"
+ */
+ virtual std::string to_string(const Corpus2::Tagset& tagset) const;
+
+ /**
+ * String representation of conditional operator in form of:
+ * "upper(strset_expr_raw_string)"
+ * This version does not require tagset, but may be inclomplete
+ * and/or contain internal info.
+ */
+ virtual std::string to_raw_string() const;
+
+ virtual const std::string raw_operator_name() const {
+ return "upper";
+ }
+
+protected:
+ const StrSetFunctionPtr strset_expr_;
+
+ typedef FunctionBase::BaseRetValPtr BaseRetValPtr;
+
+ /**
+ * Get a string set from the argument expression and return copy of the set
+ * with all strings in upper case form
+ */
+ virtual BaseRetValPtr apply_internal(const SentenceContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_TOUPPER_H
diff --git a/tests/strsetfunctions.cpp b/tests/strsetfunctions.cpp
index 4822967..cd64290 100644
--- a/tests/strsetfunctions.cpp
+++ b/tests/strsetfunctions.cpp
@@ -7,6 +7,7 @@
#include <libwccl/values/strset.h>
#include <libwccl/sentencecontext.h>
#include <libwccl/ops/tolower.h>
+#include <libwccl/ops/toupper.h>
#include <libwccl/ops/constant.h>
using namespace Wccl;
@@ -77,6 +78,28 @@ BOOST_FIXTURE_TEST_CASE(lower_locale, StrSetFix)
BOOST_CHECK(lowerset.equals(*to_lower.apply(sc)));
}
+BOOST_FIXTURE_TEST_CASE(upper_locale, StrSetFix)
+{
+ //I'm not sure if I can guarantee this test will pass
+ //on all locales? - ToUpper uses default locale at the moment
+
+ //I wanted to make sure switching around encoding of source file
+ //won't affect the test, so I explicitly provide escaped UTF8 sequence
+
+ StrSet upperset;
+ upperset.insert(UnicodeString::fromUTF8(
+ "ZA\xC5\xBB\xC3\x93\xC5\x81\xC4\x86G\xC4\x98\xC5\x9AL\xC4\x84JA\xC5\xB9\xC5\x83"
+ "ZA\xC5\xBB\xC3\x93\xC5\x81\xC4\x86G\xC4\x98\xC5\x9AL\xC4\x84JA\xC5\xB9\xC5\x83"));
+ StrSet lowerset;
+ lowerset.insert(UnicodeString::fromUTF8(
+ "za\xC5\xBC\xC3\xB3\xC5\x82\xC4\x87g\xC4\x99\xC5\x9Bl\xC4\x85ja\xC5\xBA\xC5\x84"
+ "ZA\xC5\xBB\xC3\x93\xC5\x81\xC4\x86g\xC4\x98\xC5\x9AL\xC4\x84JA\xC5\xB9\xC5\x83"));
+
+ ToUpper to_upper(boost::shared_ptr<Function<StrSet> >(
+ new Constant<StrSet>(lowerset)));
+
+ BOOST_CHECK(upperset.equals(*to_upper.apply(sc)));
+}
//------ to_string test cases -------
@@ -90,4 +113,34 @@ BOOST_FIXTURE_TEST_CASE(lower_to_string, StrSetFix)
BOOST_CHECK_EQUAL(expected, to_lower.to_string(tagset));
}
+BOOST_AUTO_TEST_CASE(lower_to_raw_string)
+{
+ StrSet one_elem_set;
+ one_elem_set.insert("YayaAy");
+ ToLower to_lower(boost::shared_ptr<Function<StrSet> >(
+ new Constant<StrSet>(one_elem_set)));
+ std::string expected = "lower([\"YayaAy\"])";
+ BOOST_CHECK_EQUAL(expected, to_lower.to_raw_string());
+}
+
+BOOST_FIXTURE_TEST_CASE(upper_to_string, StrSetFix)
+{
+ StrSet one_elem_set;
+ one_elem_set.insert("YayaAy");
+ ToUpper to_upper(boost::shared_ptr<Function<StrSet> >(
+ new Constant<StrSet>(one_elem_set)));
+ std::string expected = "upper([\"YayaAy\"])";
+ BOOST_CHECK_EQUAL(expected, to_upper.to_string(tagset));
+}
+
+BOOST_AUTO_TEST_CASE(upper_to_raw_string)
+{
+ StrSet one_elem_set;
+ one_elem_set.insert("YayaAy");
+ ToUpper to_upper(boost::shared_ptr<Function<StrSet> >(
+ new Constant<StrSet>(one_elem_set)));
+ std::string expected = "upper([\"YayaAy\"])";
+ BOOST_CHECK_EQUAL(expected, to_upper.to_raw_string());
+}
+
BOOST_AUTO_TEST_SUITE_END()
--
GitLab