Implement TextServer `strip_diacritics` function.

2 files changed, 36 insertions, 0 deletions

diff --git a/modules/text_server_adv/text_server_adv.cpp b/modules/text_server_adv/text_server_adv.cpp
index c459141265..776134b598 100644
--- a/modules/text_server_adv/text_server_adv.cpp
+++ b/modules/text_server_adv/text_server_adv.cpp
@@ -4924,6 +4924,39 @@ String TextServerAdvanced::percent_sign(const String &p_language) const {
 	return "%";
 }
 
+String TextServerAdvanced::strip_diacritics(const String &p_string) const {
+	UErrorCode err = U_ZERO_ERROR;
+
+	// Get NFKD normalizer singleton.
+	const UNormalizer2 *unorm = unorm2_getNFKDInstance(&err);
+	ERR_FAIL_COND_V_MSG(U_FAILURE(err), TextServer::strip_diacritics(p_string), u_errorName(err));
+
+	// Convert to UTF-16.
+	Char16String utf16 = p_string.utf16();
+
+	// Normalize.
+	Char16String normalized;
+	err = U_ZERO_ERROR;
+	int32_t len = unorm2_normalize(unorm, utf16.ptr(), -1, nullptr, 0, &err);
+	ERR_FAIL_COND_V_MSG(err != U_BUFFER_OVERFLOW_ERROR, TextServer::strip_diacritics(p_string), u_errorName(err));
+	normalized.resize(len);
+	err = U_ZERO_ERROR;
+	unorm2_normalize(unorm, utf16.ptr(), -1, normalized.ptrw(), len, &err);
+	ERR_FAIL_COND_V_MSG(U_FAILURE(err), TextServer::strip_diacritics(p_string), u_errorName(err));
+
+	// Convert back to UTF-32.
+	String normalized_string = String::utf16(normalized.ptr(), len);
+
+	// Strip combining characters.
+	String result;
+	for (int i = 0; i < normalized_string.length(); i++) {
+		if (u_getCombiningClass(normalized_string[i]) == 0) {
+			result += normalized_string[i];
+		}
+	}
+	return result;
+}
+
 TextServerAdvanced::TextServerAdvanced() {
 	_insert_num_systems_lang();
 	_insert_feature_sets();
diff --git a/modules/text_server_adv/text_server_adv.h b/modules/text_server_adv/text_server_adv.h
index 333b68e074..15f3a7f1a9 100644
--- a/modules/text_server_adv/text_server_adv.h
+++ b/modules/text_server_adv/text_server_adv.h
@@ -50,6 +50,7 @@
 #include <unicode/udata.h>
 #include <unicode/uiter.h>
 #include <unicode/uloc.h>
+#include <unicode/unorm2.h>
 #include <unicode/uscript.h>
 #include <unicode/ustring.h>
 #include <unicode/utypes.h>
@@ -501,6 +502,8 @@ public:
 	virtual String parse_number(const String &p_string, const String &p_language = "") const override;
 	virtual String percent_sign(const String &p_language = "") const override;
 
+	virtual String strip_diacritics(const String &p_string) const override;
+
 	TextServerAdvanced();
 	~TextServerAdvanced();
 };