4 files changed, 58 insertions, 35 deletions
diff --git a/indra/llcommon/CMakeLists.txt b/indra/llcommon/CMakeLists.txt
index c947184dc8..5f4ed2fffa 100644
--- a/indra/llcommon/CMakeLists.txt
+++ b/indra/llcommon/CMakeLists.txt
@@ -3,7 +3,6 @@
 project(llcommon)
 
 include(00-Common)
-include(ICU4C)
 include(LLCommon)
 include(bugsplat)
 include(Linking)
@@ -283,7 +282,6 @@ target_link_libraries(
         ll::uriparser
         ll::oslibraries
         ll::tracy
-        ll::icu4c
     )
 
 target_include_directories(llcommon INTERFACE ${CMAKE_CURRENT_SOURCE_DIR})
diff --git a/indra/llcommon/llstring.cpp b/indra/llcommon/llstring.cpp
index a743ae1589..6512bbc392 100644
--- a/indra/llcommon/llstring.cpp
+++ b/indra/llcommon/llstring.cpp
@@ -30,7 +30,6 @@
 #include "llerror.h"
 #include "llfasttimer.h"
 #include "llsd.h"
-#include <unicode/uchar.h>
 #include <vector>
 
 #if LL_WINDOWS
@@ -758,6 +757,43 @@ std::string utf8str_showBytesUTF8(const std::string& utf8str)
     return result;
 }
 
+// Search for any emoji symbol, return true if found
+bool wstring_has_emoji(const LLWString& wstr)
+{
+    for (const llwchar& wch : wstr)
+    {
+        if (LLStringOps::isEmoji(wch))
+            return true;
+    }
+
+    return false;
+}
+
+// Cut emoji symbols if exist
+bool wstring_remove_emojis(LLWString& wstr)
+{
+    bool found = false;
+    for (size_t i = 0; i < wstr.size(); ++i)
+    {
+        if (LLStringOps::isEmoji(wstr[i]))
+        {
+            wstr.erase(i--, 1);
+            found = true;
+        }
+    }
+    return found;
+}
+
+// Cut emoji symbols if exist
+bool utf8str_remove_emojis(std::string& utf8str)
+{
+    LLWString wstr = utf8str_to_wstring(utf8str);
+    if (!wstring_remove_emojis(wstr))
+        return false;
+    utf8str = wstring_to_utf8str(wstr);
+    return true;
+}
+
 #if LL_WINDOWS
 unsigned int ll_wstring_default_code_page()
 {
@@ -971,40 +1007,18 @@ std::string LLStringOps::sAM;
 std::string LLStringOps::sPM;
 
 // static
-bool LLStringOps::isEmoji(llwchar wch)
-{
-    int ublock = ublock_getCode(wch);
-    switch (ublock)
-    {
-        case UBLOCK_GENERAL_PUNCTUATION:
-        case UBLOCK_LETTERLIKE_SYMBOLS:
-        case UBLOCK_ARROWS:
-        case UBLOCK_MISCELLANEOUS_TECHNICAL:
-        case UBLOCK_ENCLOSED_ALPHANUMERICS:
-        case UBLOCK_GEOMETRIC_SHAPES:
-        case UBLOCK_MISCELLANEOUS_SYMBOLS:
-        case UBLOCK_DINGBATS:
-        case UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION:
-        case UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS:
-        case UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS:
-        case UBLOCK_EMOTICONS:
-        case UBLOCK_TRANSPORT_AND_MAP_SYMBOLS:
-#if U_ICU_VERSION_MAJOR_NUM > 56
-        // Boost uses ICU so we can't update it independently
-        case UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS:
-#endif // U_ICU_VERSION_MAJOR_NUM > 56
-            return true;
-        default:
-#if U_ICU_VERSION_MAJOR_NUM > 56
-            return false;
+bool LLStringOps::isEmoji(llwchar a)
+{
+#if 0   // Do not consider special characters that might have a corresponding
+        // glyph in the monochorme fallback fonts as a "genuine" emoji. HB
+    return a == 0xa9 || a == 0xae || (a >= 0x2000 && a < 0x3300) ||
+           (a >= 0x1f000 && a < 0x20000);
 #else
-            // See https://en.wikipedia.org/wiki/Supplemental_Symbols_and_Pictographs
-            return wch >= 0x1F900 && wch <= 0x1F9FF;
-#endif // U_ICU_VERSION_MAJOR_NUM > 56
-    }
+    // These are indeed "genuine" emojis, we *do want* rendered as such. HB
+    return a >= 0x1f000 && a < 0x20000;
+#endif
 }
 
-
 S32 LLStringOps::collate(const llwchar* a, const llwchar* b)
 {
     #if LL_WINDOWS
diff --git a/indra/llcommon/llstring.h b/indra/llcommon/llstring.h
index 3829978ddf..6503da2e77 100644
--- a/indra/llcommon/llstring.h
+++ b/indra/llcommon/llstring.h
@@ -189,7 +189,8 @@ public:
     static bool isAlnum(char a) { return isalnum((unsigned char)a) != 0; }
     static bool isAlnum(llwchar a) { return iswalnum(a) != 0; }
 
-    static bool isEmoji(llwchar wch);
+    // Returns true when 'a' corresponds to a "genuine" emoji. HB
+    static bool isEmoji(llwchar a);
 
     static S32  collate(const char* a, const char* b) { return strcoll(a, b); }
     static S32  collate(const llwchar* a, const llwchar* b);
@@ -747,6 +748,12 @@ LL_COMMON_API llwchar utf8str_to_wchar(const std::string& utf8str, size_t offset
 
 LL_COMMON_API std::string utf8str_showBytesUTF8(const std::string& utf8str);
 
+LL_COMMON_API bool wstring_has_emoji(const LLWString& wstr);
+
+LL_COMMON_API bool wstring_remove_emojis(LLWString& wstr);
+
+LL_COMMON_API bool utf8str_remove_emojis(std::string& utf8str);
+
 #if LL_WINDOWS
 /* @name Windows string helpers
  */
diff --git a/indra/llcommon/llsys.cpp b/indra/llcommon/llsys.cpp
index 47868b3fca..b7ffddc023 100644
--- a/indra/llcommon/llsys.cpp
+++ b/indra/llcommon/llsys.cpp
@@ -1350,6 +1350,10 @@ BOOL gunzip_file(const std::string& srcfile, const std::string& dstfile)
     } while(gzeof(src) == 0);
     fclose(dst);
     dst = NULL;
+#if LL_WINDOWS
+    // Rename in windows needs the dstfile to not exist.
+    LLFile::remove(dstfile, ENOENT);
+#endif
     if (LLFile::rename(tmpfile, dstfile) == -1) goto err;       /* Flawfinder: ignore */
     retval = TRUE;
 err: