Add Collapse variant for utf-32 strings

ref:5897581c4d2be8d8775525221d2d910c04d788b8

3 files changed, 32 insertions, 5 deletions

diff --git a/util/string/strip.cpp b/util/string/strip.cpp
index c921571cf0..1cf5e33096 100644
--- a/util/string/strip.cpp
+++ b/util/string/strip.cpp
@@ -3,10 +3,6 @@
 
 #include <util/string/reverse.h>
 
-bool Collapse(const TString& from, TString& to, size_t maxLen) {
-    return CollapseImpl<TString, bool (*)(unsigned char)>(from, to, maxLen, IsAsciiSpace);
-}
-
 void CollapseText(const TString& from, TString& to, size_t maxLen) {
     Collapse(from, to, maxLen);
     StripInPlace(to);
diff --git a/util/string/strip.h b/util/string/strip.h
index d5ef6da96d..a65d3fe069 100644
--- a/util/string/strip.h
+++ b/util/string/strip.h
@@ -230,13 +230,27 @@ bool CollapseImpl(const TStringType& from, TStringType& to, size_t maxLen, const
     return false;
 }
 
-bool Collapse(const TString& from, TString& to, size_t maxLen = 0);
+template <class TStringType, class TWhitespaceFunc>
+std::enable_if_t<std::is_invocable_v<TWhitespaceFunc, typename TStringType::value_type>, bool> Collapse(
+    const TStringType& from, TStringType& to, TWhitespaceFunc isWhitespace, size_t maxLen = 0)
+{
+    return CollapseImpl(from, to, maxLen, isWhitespace);
+}
+
+inline bool Collapse(const TString& from, TString& to, size_t maxLen = 0) {
+    return Collapse(from, to, IsAsciiSpace<typename TString::value_type>, maxLen);
+}
 
 /// Replaces several consequtive space symbols with one (processing is limited to maxLen bytes)
 inline TString& CollapseInPlace(TString& s, size_t maxLen = 0) {
     Collapse(s, s, maxLen);
     return s;
 }
+template <class TStringType, class TWhitespaceFunc>
+inline TStringType& CollapseInPlace(TStringType& s, TWhitespaceFunc isWhitespace, size_t maxLen = 0) {
+    Collapse(s, s, isWhitespace, maxLen);
+    return s;
+}
 
 /// Replaces several consequtive space symbols with one (processing is limited to maxLen bytes)
 inline TString Collapse(const TString& s, size_t maxLen = 0) Y_WARN_UNUSED_RESULT;
diff --git a/util/string/strip_ut.cpp b/util/string/strip_ut.cpp
index d1029d1498..df4f9bc57d 100644
--- a/util/string/strip_ut.cpp
+++ b/util/string/strip_ut.cpp
@@ -107,6 +107,23 @@ Y_UNIT_TEST_SUITE(TStripStringTest) {
             u"abc");
     }
 
+    Y_UNIT_TEST(TestCollapseUtf32) {
+        TUtf32String s;
+        Collapse(UTF8ToUTF32<true>("  123    456  "), s, IsWhitespace);
+        UNIT_ASSERT(s == UTF8ToUTF32<true>(" 123 456 "));
+        Collapse(UTF8ToUTF32<true>("  123    456  "), s, IsWhitespace, 10);
+        UNIT_ASSERT(s == UTF8ToUTF32<true>(" 123 456  "));
+
+        s = UTF8ToUTF32<true>(" a b c ");
+        TUtf32String s2 = s;
+        CollapseInPlace(s2, IsWhitespace);
+
+        UNIT_ASSERT(s == s2);
+#ifndef TSTRING_IS_STD_STRING
+        UNIT_ASSERT(s.c_str() == s2.c_str()); // Collapse() does not change the string at all
+#endif
+    }
+
     Y_UNIT_TEST(TestCollapse) {
         TString s;
         Collapse(TString("  123    456  "), s);