From f6a79879005ff80b7aaa814b6c2b2f271f6b1169 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Adam=20Wardy=C5=84ski?= <no@email>
Date: Thu, 11 Nov 2010 22:44:10 +0100
Subject: [PATCH] Affix operator (returning prefixes or suffixes of given
length)
---
libwccl/CMakeLists.txt | 3 +-
libwccl/ops/affix.cpp | 44 ++++++++++++++++++++++++++++
libwccl/ops/affix.h | 59 ++++++++++++++++++++++++++++++++++++++
tests/strsetfunctions.cpp | 60 +++++++++++++++++++++++++++++++++++++++
4 files changed, 165 insertions(+), 1 deletion(-)
create mode 100644 libwccl/ops/affix.cpp
create mode 100644 libwccl/ops/affix.h
diff --git a/libwccl/CMakeLists.txt b/libwccl/CMakeLists.txt
index c5fb97c..1a86bd6 100644
--- a/libwccl/CMakeLists.txt
+++ b/libwccl/CMakeLists.txt
@@ -16,9 +16,10 @@ set(LIBS ${LIBS} ${Boost_LIBRARIES})
SET(libwccl_STAT_SRC
exception.cpp
ops/and.cpp
+ ops/affix.cpp
ops/formatters.cpp
ops/logicalpredicate.cpp
- ops/nor.cpp
+ ops/nor.cpp
ops/or.cpp
ops/predicate.cpp
ops/tolower.cpp
diff --git a/libwccl/ops/affix.cpp b/libwccl/ops/affix.cpp
new file mode 100644
index 0000000..892a3c0
--- /dev/null
+++ b/libwccl/ops/affix.cpp
@@ -0,0 +1,44 @@
+#include <libwccl/ops/affix.h>
+#include <sstream>
+#include <boost/foreach.hpp>
+#define foreach BOOST_FOREACH
+
+namespace Wccl {
+
+std::string Affix::to_string(const Corpus2::Tagset& tagset) const
+{
+ std::stringstream str;
+ str << operator_name(tagset) << "(" << strset_expr_->to_string(tagset)
+ << ", " << affix_length_ << ")";
+ return str.str();
+}
+
+std::string Affix::to_raw_string() const {
+ std::stringstream str;
+ str << raw_operator_name() << "(" << strset_expr_->to_raw_string()
+ << ", " << affix_length_ << ")";
+ return str.str();
+}
+
+Affix::BaseRetValPtr Affix::apply_internal(const SentenceContext& context) const
+{
+ if(affix_length_ == 0) {
+ return strset_expr_->apply(context);
+ }
+ const boost::shared_ptr<StrSet>& set = strset_expr_->apply(context);
+ boost::shared_ptr<StrSet> a_set = boost::shared_ptr<StrSet>(new StrSet());
+ if(affix_length_ < 0) {
+ foreach(const UnicodeString& s, set->contents()) {
+ a_set->insert(UnicodeString(s).remove(0, s.length() + affix_length_));
+ }
+ } else {
+ foreach(const UnicodeString& s, set->contents()) {
+ UnicodeString prefixed(s);
+ prefixed.truncate(affix_length_);
+ a_set->insert(prefixed);
+ }
+ }
+ return a_set;
+}
+
+} /* end ns Wccl */
diff --git a/libwccl/ops/affix.h b/libwccl/ops/affix.h
new file mode 100644
index 0000000..fc38e59
--- /dev/null
+++ b/libwccl/ops/affix.h
@@ -0,0 +1,59 @@
+#ifndef LIBWCCL_OPS_AFFIX_H
+#define LIBWCCL_OPS_AFFIX_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 that are prefixes or
+ * suffixes of given length
+ */
+class Affix : public Function<StrSet> {
+public:
+ typedef boost::shared_ptr<Function<StrSet> > StrSetFunctionPtr;
+
+ Affix(const StrSetFunctionPtr& strset_expr, int affix_length)
+ : strset_expr_(strset_expr),
+ affix_length_(affix_length)
+ {
+ BOOST_ASSERT(strset_expr_);
+ }
+
+ /**
+ * String representation of the operator in form of:
+ * "affix(strset_expr_string)"
+ */
+ virtual std::string to_string(const Corpus2::Tagset& tagset) const;
+
+ /**
+ * String representation of conditional operator in form of:
+ * "affix(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 "affix";
+ }
+
+protected:
+ const StrSetFunctionPtr strset_expr_;
+ const int affix_length_;
+
+ typedef FunctionBase::BaseRetValPtr BaseRetValPtr;
+
+ /**
+ * Get a string set from the argument expression and return copy of the set
+ * with all strings converted into prefixes or suffixes of given length
+ */
+ virtual BaseRetValPtr apply_internal(const SentenceContext& context) const;
+};
+
+} /* end ns Wccl */
+
+#endif // LIBWCCL_OPS_AFFIX_H
diff --git a/tests/strsetfunctions.cpp b/tests/strsetfunctions.cpp
index 337f36c..bf0ecc9 100644
--- a/tests/strsetfunctions.cpp
+++ b/tests/strsetfunctions.cpp
@@ -8,6 +8,7 @@
#include <libwccl/sentencecontext.h>
#include <libwccl/ops/tolower.h>
#include <libwccl/ops/toupper.h>
+#include <libwccl/ops/affix.h>
#include <libwccl/ops/constant.h>
using namespace Wccl;
@@ -29,6 +30,7 @@ struct StrSetFix
strset.insert("some1325numbers");
strset.insert("ALLUPPER");
strset.insert("kIdSpEeChLoL");
+ strset.insert("short");
strset_expr.reset(new Constant<StrSet>(strset));
}
@@ -49,12 +51,70 @@ BOOST_FIXTURE_TEST_CASE(lower, StrSetFix)
lowerset.insert("some1325numbers");
lowerset.insert("allupper");
lowerset.insert("kidspeechlol");
+ lowerset.insert("short");
ToLower to_lower(strset_expr);
BOOST_CHECK(lowerset.equals(*to_lower.apply(sc)));
}
+BOOST_FIXTURE_TEST_CASE(upper, StrSetFix)
+{
+ StrSet upperset;
+ upperset.insert("ALLLOWER");
+ upperset.insert("FIRSTCAPITAL");
+ upperset.insert("PASCALCASE");
+ upperset.insert("CAMELCASE");
+ upperset.insert("SOME1325NUMBERS");
+ upperset.insert("ALLUPPER");
+ upperset.insert("KIDSPEECHLOL");
+ upperset.insert("SHORT");
+
+ ToUpper to_upper(strset_expr);
+
+ BOOST_CHECK(upperset.equals(*to_upper.apply(sc)));
+}
+
+BOOST_FIXTURE_TEST_CASE(prefix, StrSetFix)
+{
+ StrSet prefixset;
+ prefixset.insert("alllowe");
+ prefixset.insert("Firstca");
+ prefixset.insert("PascalC");
+ prefixset.insert("camelCa");
+ prefixset.insert("some132");
+ prefixset.insert("ALLUPPE");
+ prefixset.insert("kIdSpEe");
+ prefixset.insert("short");
+
+ Affix prefix(strset_expr, 7);
+
+ BOOST_CHECK(prefixset.equals(*prefix.apply(sc)));
+}
+
+BOOST_FIXTURE_TEST_CASE(suffix, StrSetFix)
+{
+ StrSet suffixset;
+ suffixset.insert("lllower");
+ suffixset.insert("capital");
+ suffixset.insert("calCase");
+ suffixset.insert("melCase");
+ suffixset.insert("numbers");
+ suffixset.insert("LLUPPER");
+ suffixset.insert("EeChLoL");
+ suffixset.insert("short");
+
+ Affix suffix(strset_expr, -7);
+
+ BOOST_CHECK(suffixset.equals(*suffix.apply(sc)));
+}
+
+BOOST_FIXTURE_TEST_CASE(affix_0, StrSetFix)
+{
+ Affix affix_0(strset_expr, 0);
+ BOOST_CHECK(strset.equals(*affix_0.apply(sc)));
+}
+
BOOST_FIXTURE_TEST_CASE(lower_locale, StrSetFix)
{
//I'm not sure if I can guarantee this test will pass
--
GitLab