[TextServer] Add ICU Unicode security and spoofing detection.

3 files changed, 75 insertions, 3 deletions

diff --git a/modules/text_server_adv/SCsub b/modules/text_server_adv/SCsub
index d212fe62b4..73e5c2bf74 100644
--- a/modules/text_server_adv/SCsub
+++ b/modules/text_server_adv/SCsub
@@ -121,7 +121,7 @@ if env["builtin_harfbuzz"]:
 
     env_harfbuzz.Append(CCFLAGS=["-DHAVE_ICU"])
     if env["builtin_icu"]:
-        env_harfbuzz.Prepend(CPPPATH=["#thirdparty/icu4c/common/"])
+        env_harfbuzz.Prepend(CPPPATH=["#thirdparty/icu4c/common/", "#thirdparty/icu4c/i18n/"])
         env_harfbuzz.Append(CCFLAGS=["-DU_HAVE_LIB_SUFFIX=1", "-DU_LIB_SUFFIX_C_NAME=_godot", "-DHAVE_ICU_BUILTIN"])
 
     if freetype_enabled:
@@ -439,6 +439,10 @@ if env["builtin_icu"]:
         "common/uvectr32.cpp",
         "common/uvectr64.cpp",
         "common/wintz.cpp",
+        "i18n/scriptset.cpp",
+        "i18n/ucln_in.cpp",
+        "i18n/uspoof.cpp",
+        "i18n/uspoof_impl.cpp",
     ]
     thirdparty_sources = [thirdparty_dir + file for file in thirdparty_sources]
 
@@ -451,7 +455,7 @@ if env["builtin_icu"]:
     else:
         thirdparty_sources += ["icu_data/icudata_stub.cpp"]
 
-    env_icu.Prepend(CPPPATH=["#thirdparty/icu4c/common/"])
+    env_icu.Prepend(CPPPATH=["#thirdparty/icu4c/common/", "#thirdparty/icu4c/i18n/"])
     env_icu.Append(
         CXXFLAGS=[
             "-DU_STATIC_IMPLEMENTATION",
@@ -463,6 +467,7 @@ if env["builtin_icu"]:
             "-DUCONFIG_NO_IDNA",
             "-DUCONFIG_NO_FILE_IO",
             "-DUCONFIG_NO_TRANSLITERATION",
+            "-DUCONFIG_NO_REGULAR_EXPRESSIONS",
             "-DPKGDATA_MODE=static",
             "-DU_ENABLE_DYLOAD=0",
             "-DU_HAVE_LIB_SUFFIX=1",
@@ -480,7 +485,7 @@ if env["builtin_icu"]:
     if env_text_server_adv["tools"]:
         env_text_server_adv.Append(CXXFLAGS=["-DICU_STATIC_DATA"])
 
-    env_text_server_adv.Prepend(CPPPATH=["#thirdparty/icu4c/common/"])
+    env_text_server_adv.Prepend(CPPPATH=["#thirdparty/icu4c/common/", "#thirdparty/icu4c/i18n/"])
 
     lib = env_icu.add_library("icu_builtin", thirdparty_sources)
     thirdparty_obj += lib
diff --git a/modules/text_server_adv/text_server_adv.cpp b/modules/text_server_adv/text_server_adv.cpp
index f8dbbc2e61..bb49fb5248 100644
--- a/modules/text_server_adv/text_server_adv.cpp
+++ b/modules/text_server_adv/text_server_adv.cpp
@@ -347,6 +347,7 @@ bool TextServerAdvanced::has_feature(Feature p_feature) const {
 		case FEATURE_CONTEXT_SENSITIVE_CASE_CONVERSION:
 		case FEATURE_USE_SUPPORT_DATA:
 		case FEATURE_UNICODE_IDENTIFIERS:
+		case FEATURE_UNICODE_SECURITY:
 			return true;
 		default: {
 		}
@@ -5640,6 +5641,68 @@ String TextServerAdvanced::percent_sign(const String &p_language) const {
 	return "%";
 }
 
+int TextServerAdvanced::is_confusable(const String &p_string, const PackedStringArray &p_dict) const {
+	UErrorCode status = U_ZERO_ERROR;
+	int match_index = -1;
+
+	Char16String utf16 = p_string.utf16();
+	Vector<UChar *> skeletons;
+	skeletons.resize(p_dict.size());
+
+	USpoofChecker *sc = uspoof_open(&status);
+	uspoof_setChecks(sc, USPOOF_CONFUSABLE, &status);
+	for (int i = 0; i < p_dict.size(); i++) {
+		Char16String word = p_dict[i].utf16();
+		int32_t len = uspoof_getSkeleton(sc, 0, word.get_data(), -1, NULL, 0, &status);
+		skeletons.write[i] = (UChar *)memalloc(++len * sizeof(UChar));
+		status = U_ZERO_ERROR;
+		uspoof_getSkeleton(sc, 0, word.get_data(), -1, skeletons.write[i], len, &status);
+	}
+
+	int32_t len = uspoof_getSkeleton(sc, 0, utf16.get_data(), -1, NULL, 0, &status);
+	UChar *skel = (UChar *)memalloc(++len * sizeof(UChar));
+	status = U_ZERO_ERROR;
+	uspoof_getSkeleton(sc, 0, utf16.get_data(), -1, skel, len, &status);
+	for (int i = 0; i < skeletons.size(); i++) {
+		if (u_strcmp(skel, skeletons[i]) == 0) {
+			match_index = i;
+			break;
+		}
+	}
+	memfree(skel);
+
+	for (int i = 0; i < skeletons.size(); i++) {
+		memfree(skeletons.write[i]);
+	}
+	uspoof_close(sc);
+
+	ERR_FAIL_COND_V_MSG(U_FAILURE(status), -1, u_errorName(status));
+
+	return match_index;
+}
+
+bool TextServerAdvanced::spoof_check(const String &p_string) const {
+	UErrorCode status = U_ZERO_ERROR;
+	Char16String utf16 = p_string.utf16();
+
+	USet *allowed = uset_openEmpty();
+	uset_addAll(allowed, uspoof_getRecommendedSet(&status));
+	uset_addAll(allowed, uspoof_getInclusionSet(&status));
+
+	USpoofChecker *sc = uspoof_open(&status);
+	uspoof_setAllowedChars(sc, allowed, &status);
+	uspoof_setRestrictionLevel(sc, USPOOF_MODERATELY_RESTRICTIVE);
+
+	int32_t bitmask = uspoof_check(sc, utf16.get_data(), -1, NULL, &status);
+
+	uspoof_close(sc);
+	uset_close(allowed);
+
+	ERR_FAIL_COND_V_MSG(U_FAILURE(status), false, u_errorName(status));
+
+	return (bitmask != 0);
+}
+
 String TextServerAdvanced::strip_diacritics(const String &p_string) const {
 	UErrorCode err = U_ZERO_ERROR;
 
diff --git a/modules/text_server_adv/text_server_adv.h b/modules/text_server_adv/text_server_adv.h
index 170d92f8fc..b337abea7a 100644
--- a/modules/text_server_adv/text_server_adv.h
+++ b/modules/text_server_adv/text_server_adv.h
@@ -101,6 +101,7 @@ using namespace godot;
 #include <unicode/uloc.h>
 #include <unicode/unorm2.h>
 #include <unicode/uscript.h>
+#include <unicode/uspoof.h>
 #include <unicode/ustring.h>
 #include <unicode/utypes.h>
 
@@ -701,6 +702,9 @@ public:
 
 	virtual PackedInt32Array string_get_word_breaks(const String &p_string, const String &p_language = "") const override;
 
+	virtual int is_confusable(const String &p_string, const PackedStringArray &p_dict) const override;
+	virtual bool spoof_check(const String &p_string) const override;
+
 	virtual String strip_diacritics(const String &p_string) const override;
 	virtual bool is_valid_identifier(const String &p_string) const override;