Merge pull request #41100 from bruvzg/ctl_text_server_interface

[Complex Text Layouts] Implement TextServer interface.

7 files changed, 138 insertions, 15 deletions

diff --git a/platform/android/detect.py b/platform/android/detect.py
index 0accacb679..60d4146712 100644
--- a/platform/android/detect.py
+++ b/platform/android/detect.py
@@ -215,7 +215,7 @@ def configure(env):
     env.Append(CPPFLAGS=["-isystem", env["ANDROID_NDK_ROOT"] + "/sources/cxx-stl/llvm-libc++abi/include"])
 
     # Disable exceptions and rtti on non-tools (template) builds
-    if env["tools"]:
+    if env["tools"] or env["builtin_icu"]:
         env.Append(CXXFLAGS=["-frtti"])
     else:
         env.Append(CXXFLAGS=["-fno-rtti", "-fno-exceptions"])
diff --git a/platform/android/display_server_android.cpp b/platform/android/display_server_android.cpp
index e82a12ece5..aab5da4fde 100644
--- a/platform/android/display_server_android.cpp
+++ b/platform/android/display_server_android.cpp
@@ -498,9 +498,28 @@ void DisplayServerAndroid::_set_key_modifier_state(Ref<InputEventWithModifiers>
 }
 
 void DisplayServerAndroid::process_key_event(int p_keycode, int p_scancode, int p_unicode_char, bool p_pressed) {
+	static char32_t prev_wc = 0;
+	char32_t unicode = p_unicode_char;
+	if ((p_unicode_char & 0xfffffc00) == 0xd800) {
+		if (prev_wc != 0) {
+			ERR_PRINT("invalid utf16 surrogate input");
+		}
+		prev_wc = unicode;
+		return; // Skip surrogate.
+	} else if ((unicode & 0xfffffc00) == 0xdc00) {
+		if (prev_wc == 0) {
+			ERR_PRINT("invalid utf16 surrogate input");
+			return; // Skip invalid surrogate.
+		}
+		unicode = (prev_wc << 10UL) + unicode - ((0xd800 << 10UL) + 0xdc00 - 0x10000);
+		prev_wc = 0;
+	} else {
+		prev_wc = 0;
+	}
+
 	Ref<InputEventKey> ev;
 	ev.instance();
-	int val = p_unicode_char;
+	int val = unicode;
 	int keycode = android_get_keysym(p_keycode);
 	int phy_keycode = android_get_keysym(p_scancode);
 
diff --git a/platform/javascript/detect.py b/platform/javascript/detect.py
index 71189cf697..9f584d0899 100644
--- a/platform/javascript/detect.py
+++ b/platform/javascript/detect.py
@@ -71,6 +71,8 @@ def configure(env):
             )
         # Tools need more memory. Initial stack memory in bytes. See `src/settings.js` in emscripten repository (will be renamed to INITIAL_MEMORY).
         env.Append(LINKFLAGS=["-s", "TOTAL_MEMORY=33554432"])
+    elif env["builtin_icu"]:
+        env.Append(CCFLAGS=["-frtti"])
     else:
         # Disable exceptions and rtti on non-tools (template) builds
         # These flags help keep the file size down.
diff --git a/platform/linuxbsd/detect.py b/platform/linuxbsd/detect.py
index a8bc3a09f6..277aafc107 100644
--- a/platform/linuxbsd/detect.py
+++ b/platform/linuxbsd/detect.py
@@ -205,14 +205,31 @@ def configure(env):
 
     # freetype depends on libpng and zlib, so bundling one of them while keeping others
     # as shared libraries leads to weird issues
-    if env["builtin_freetype"] or env["builtin_libpng"] or env["builtin_zlib"]:
+    if (
+        env["builtin_freetype"]
+        or env["builtin_libpng"]
+        or env["builtin_zlib"]
+        or env["builtin_graphite"]
+        or env["builtin_harfbuzz"]
+    ):
         env["builtin_freetype"] = True
         env["builtin_libpng"] = True
         env["builtin_zlib"] = True
+        env["builtin_graphite"] = True
+        env["builtin_harfbuzz"] = True
 
     if not env["builtin_freetype"]:
         env.ParseConfig("pkg-config freetype2 --cflags --libs")
 
+    if not env["builtin_graphite"]:
+        env.ParseConfig("pkg-config graphite2 --cflags --libs")
+
+    if not env["builtin_icu"]:
+        env.ParseConfig("pkg-config icu-uc --cflags --libs")
+
+    if not env["builtin_harfbuzz"]:
+        env.ParseConfig("pkg-config harfbuzz harfbuzz-icu --cflags --libs")
+
     if not env["builtin_libpng"]:
         env.ParseConfig("pkg-config libpng16 --cflags --libs")
 
diff --git a/platform/osx/display_server_osx.mm b/platform/osx/display_server_osx.mm
index 1ad7117b39..3025210195 100644
--- a/platform/osx/display_server_osx.mm
+++ b/platform/osx/display_server_osx.mm
@@ -701,8 +701,6 @@ static const NSRange kEmptyRange = { NSNotFound, 0 };
 		characters = (NSString *)aString;
 	}
 
-	NSUInteger i, length = [characters length];
-
 	NSCharacterSet *ctrlChars = [NSCharacterSet controlCharacterSet];
 	NSCharacterSet *wsnlChars = [NSCharacterSet whitespaceAndNewlineCharacterSet];
 	if ([characters rangeOfCharacterFromSet:ctrlChars].length && [characters rangeOfCharacterFromSet:wsnlChars].length == 0) {
@@ -712,8 +710,15 @@ static const NSRange kEmptyRange = { NSNotFound, 0 };
 		return;
 	}
 
-	for (i = 0; i < length; i++) {
-		const unichar codepoint = [characters characterAtIndex:i];
+	Char16String text;
+	text.resize([characters length] + 1);
+	[characters getCharacters:(unichar *)text.ptrw() range:NSMakeRange(0, [characters length])];
+
+	String u32text;
+	u32text.parse_utf16(text.ptr(), text.length());
+
+	for (int i = 0; i < u32text.length(); i++) {
+		const char32_t codepoint = u32text[i];
 		if ((codepoint & 0xFF00) == 0xF700) {
 			continue;
 		}
@@ -1315,7 +1320,16 @@ static int remapKey(unsigned int key, unsigned int state) {
 
 		if (!wd.im_active && length > 0 && keycode_has_unicode(remapKey([event keyCode], [event modifierFlags]))) {
 			// Fallback unicode character handler used if IME is not active
-			for (NSUInteger i = 0; i < length; i++) {
+			Char16String text;
+			text.resize([characters length] + 1);
+			[characters getCharacters:(unichar *)text.ptrw() range:NSMakeRange(0, [characters length])];
+
+			String u32text;
+			u32text.parse_utf16(text.ptr(), text.length());
+
+			for (int i = 0; i < u32text.length(); i++) {
+				const char32_t codepoint = u32text[i];
+
 				DisplayServerOSX::KeyEvent ke;
 
 				ke.window_id = window_id;
@@ -1325,7 +1339,7 @@ static int remapKey(unsigned int key, unsigned int state) {
 				ke.keycode = remapKey([event keyCode], [event modifierFlags]);
 				ke.physical_keycode = translateKey([event keyCode]);
 				ke.raw = true;
-				ke.unicode = [characters characterAtIndex:i];
+				ke.unicode = codepoint;
 
 				_push_to_key_event_buffer(ke);
 			}
@@ -1417,7 +1431,15 @@ static int remapKey(unsigned int key, unsigned int state) {
 
 		// Fallback unicode character handler used if IME is not active
 		if (!wd.im_active && length > 0 && keycode_has_unicode(remapKey([event keyCode], [event modifierFlags]))) {
-			for (NSUInteger i = 0; i < length; i++) {
+			Char16String text;
+			text.resize([characters length] + 1);
+			[characters getCharacters:(unichar *)text.ptrw() range:NSMakeRange(0, [characters length])];
+
+			String u32text;
+			u32text.parse_utf16(text.ptr(), text.length());
+
+			for (int i = 0; i < u32text.length(); i++) {
+				const char32_t codepoint = u32text[i];
 				DisplayServerOSX::KeyEvent ke;
 
 				ke.window_id = window_id;
@@ -1427,7 +1449,7 @@ static int remapKey(unsigned int key, unsigned int state) {
 				ke.keycode = remapKey([event keyCode], [event modifierFlags]);
 				ke.physical_keycode = translateKey([event keyCode]);
 				ke.raw = true;
-				ke.unicode = [characters characterAtIndex:i];
+				ke.unicode = codepoint;
 
 				_push_to_key_event_buffer(ke);
 			}
diff --git a/platform/server/detect.py b/platform/server/detect.py
index d9ac357679..bf4744a234 100644
--- a/platform/server/detect.py
+++ b/platform/server/detect.py
@@ -138,14 +138,31 @@ def configure(env):
 
     # freetype depends on libpng and zlib, so bundling one of them while keeping others
     # as shared libraries leads to weird issues
-    if env["builtin_freetype"] or env["builtin_libpng"] or env["builtin_zlib"]:
+    if (
+        env["builtin_freetype"]
+        or env["builtin_libpng"]
+        or env["builtin_zlib"]
+        or env["builtin_graphite"]
+        or env["builtin_harfbuzz"]
+    ):
         env["builtin_freetype"] = True
         env["builtin_libpng"] = True
         env["builtin_zlib"] = True
+        env["builtin_graphite"] = True
+        env["builtin_harfbuzz"] = True
 
     if not env["builtin_freetype"]:
         env.ParseConfig("pkg-config freetype2 --cflags --libs")
 
+    if not env["builtin_graphite"]:
+        env.ParseConfig("pkg-config graphite2 --cflags --libs")
+
+    if not env["builtin_icu"]:
+        env.ParseConfig("pkg-config icu-uc --cflags --libs")
+
+    if not env["builtin_harfbuzz"]:
+        env.ParseConfig("pkg-config harfbuzz harfbuzz-icu --cflags --libs")
+
     if not env["builtin_libpng"]:
         env.ParseConfig("pkg-config libpng16 --cflags --libs")
 
@@ -233,7 +250,17 @@ def configure(env):
     env.Append(CPPDEFINES=["SERVER_ENABLED", "UNIX_ENABLED"])
 
     if platform.system() == "Darwin":
-        env.Append(LINKFLAGS=["-framework", "Cocoa", "-framework", "Carbon", "-lz", "-framework", "IOKit"])
+        env.Append(
+            LINKFLAGS=[
+                "-framework",
+                "Cocoa",
+                "-framework",
+                "Carbon",
+                "-lz",
+                "-framework",
+                "IOKit",
+            ]
+        )
 
     env.Append(LIBS=["pthread"])
 
diff --git a/platform/windows/display_server_windows.cpp b/platform/windows/display_server_windows.cpp
index dfbb734ee4..2581cc3af6 100644
--- a/platform/windows/display_server_windows.cpp
+++ b/platform/windows/display_server_windows.cpp
@@ -2792,6 +2792,24 @@ void DisplayServerWindows::_process_key_events() {
 			case WM_CHAR: {
 				// extended keys should only be processed as WM_KEYDOWN message.
 				if (!KeyMappingWindows::is_extended_key(ke.wParam) && ((i == 0 && ke.uMsg == WM_CHAR) || (i > 0 && key_event_buffer[i - 1].uMsg == WM_CHAR))) {
+					static char32_t prev_wc = 0;
+					char32_t unicode = ke.wParam;
+					if ((unicode & 0xfffffc00) == 0xd800) {
+						if (prev_wc != 0) {
+							ERR_PRINT("invalid utf16 surrogate input");
+						}
+						prev_wc = unicode;
+						break; // Skip surrogate.
+					} else if ((unicode & 0xfffffc00) == 0xdc00) {
+						if (prev_wc == 0) {
+							ERR_PRINT("invalid utf16 surrogate input");
+							break; // Skip invalid surrogate.
+						}
+						unicode = (prev_wc << 10UL) + unicode - ((0xd800 << 10UL) + 0xdc00 - 0x10000);
+						prev_wc = 0;
+					} else {
+						prev_wc = 0;
+					}
 					Ref<InputEventKey> k;
 					k.instance();
 
@@ -2803,7 +2821,7 @@ void DisplayServerWindows::_process_key_events() {
 					k->set_pressed(true);
 					k->set_keycode(KeyMappingWindows::get_keysym(ke.wParam));
 					k->set_physical_keycode(KeyMappingWindows::get_scansym((ke.lParam >> 16) & 0xFF, ke.lParam & (1 << 24)));
-					k->set_unicode(ke.wParam);
+					k->set_unicode(unicode);
 					if (k->get_unicode() && gr_mem) {
 						k->set_alt(false);
 						k->set_control(false);
@@ -2840,7 +2858,25 @@ void DisplayServerWindows::_process_key_events() {
 				k->set_physical_keycode(KeyMappingWindows::get_scansym((ke.lParam >> 16) & 0xFF, ke.lParam & (1 << 24)));
 
 				if (i + 1 < key_event_pos && key_event_buffer[i + 1].uMsg == WM_CHAR) {
-					k->set_unicode(key_event_buffer[i + 1].wParam);
+					char32_t unicode = key_event_buffer[i + 1].wParam;
+					static char32_t prev_wck = 0;
+					if ((unicode & 0xfffffc00) == 0xd800) {
+						if (prev_wck != 0) {
+							ERR_PRINT("invalid utf16 surrogate input");
+						}
+						prev_wck = unicode;
+						break; // Skip surrogate.
+					} else if ((unicode & 0xfffffc00) == 0xdc00) {
+						if (prev_wck == 0) {
+							ERR_PRINT("invalid utf16 surrogate input");
+							break; // Skip invalid surrogate.
+						}
+						unicode = (prev_wck << 10UL) + unicode - ((0xd800 << 10UL) + 0xdc00 - 0x10000);
+						prev_wck = 0;
+					} else {
+						prev_wck = 0;
+					}
+					k->set_unicode(unicode);
 				}
 				if (k->get_unicode() && gr_mem) {
 					k->set_alt(false);