Implement text-to-speech support on Android, iOS, HTML5, Linux, macOS and Windows.

Implement TextServer word break method.

4 files changed, 90 insertions, 1 deletions

diff --git a/modules/text_server_adv/text_server_adv.cpp b/modules/text_server_adv/text_server_adv.cpp
index 0ae8219e23..437fbe76ab 100644
--- a/modules/text_server_adv/text_server_adv.cpp
+++ b/modules/text_server_adv/text_server_adv.cpp
@@ -3255,6 +3255,19 @@ void TextServerAdvanced::font_set_global_oversampling(double p_oversampling) {
 /* Shaped text buffer interface                                          */
 /*************************************************************************/
 
+int64_t TextServerAdvanced::_convert_pos(const String &p_utf32, const Char16String &p_utf16, int64_t p_pos) const {
+	int64_t limit = p_pos;
+	if (p_utf32.length() != p_utf16.length()) {
+		const UChar *data = p_utf16.ptr();
+		for (int i = 0; i < p_pos; i++) {
+			if (U16_IS_LEAD(data[i])) {
+				limit--;
+			}
+		}
+	}
+	return limit;
+}
+
 int64_t TextServerAdvanced::_convert_pos(const ShapedTextDataAdvanced *p_sd, int64_t p_pos) const {
 	int64_t limit = p_pos;
 	if (p_sd->text.length() != p_sd->utf16.length()) {
@@ -5555,6 +5568,53 @@ String TextServerAdvanced::string_to_lower(const String &p_string, const String
 	return String::utf16(lower.ptr(), len);
 }
 
+PackedInt32Array TextServerAdvanced::string_get_word_breaks(const String &p_string, const String &p_language) const {
+	// Convert to UTF-16.
+	Char16String utf16 = p_string.utf16();
+
+	Set<int> breaks;
+	UErrorCode err = U_ZERO_ERROR;
+	UBreakIterator *bi = ubrk_open(UBRK_LINE, p_language.ascii().get_data(), (const UChar *)utf16.ptr(), utf16.length(), &err);
+	if (U_FAILURE(err)) {
+		// No data loaded - use fallback.
+		for (int i = 0; i < p_string.length(); i++) {
+			char32_t c = p_string[i];
+			if (is_whitespace(c) || is_linebreak(c)) {
+				breaks.insert(i);
+			}
+		}
+	} else {
+		while (ubrk_next(bi) != UBRK_DONE) {
+			int pos = _convert_pos(p_string, utf16, ubrk_current(bi)) - 1;
+			if (pos != p_string.length() - 1) {
+				breaks.insert(pos);
+			}
+		}
+	}
+	ubrk_close(bi);
+
+	PackedInt32Array ret;
+	for (int i = 0; i < p_string.length(); i++) {
+		char32_t c = p_string[i];
+		if (c == 0xfffc) {
+			continue;
+		}
+		if (u_ispunct(c) && c != 0x005F) {
+			ret.push_back(i);
+			continue;
+		}
+		if (is_underscore(c)) {
+			ret.push_back(i);
+			continue;
+		}
+		if (breaks.has(i)) {
+			ret.push_back(i);
+			continue;
+		}
+	}
+	return ret;
+}
+
 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 fa59566a94..1b4293aa72 100644
--- a/modules/text_server_adv/text_server_adv.h
+++ b/modules/text_server_adv/text_server_adv.h
@@ -393,11 +393,13 @@ class TextServerAdvanced : public TextServerExtension {
 	mutable RID_PtrOwner<ShapedTextDataAdvanced> shaped_owner;
 
 	void _realign(ShapedTextDataAdvanced *p_sd) const;
+	int64_t _convert_pos(const String &p_utf32, const Char16String &p_utf16, int64_t p_pos) const;
 	int64_t _convert_pos(const ShapedTextDataAdvanced *p_sd, int64_t p_pos) const;
 	int64_t _convert_pos_inv(const ShapedTextDataAdvanced *p_sd, int64_t p_pos) const;
 	bool _shape_substr(ShapedTextDataAdvanced *p_new_sd, const ShapedTextDataAdvanced *p_sd, int64_t p_start, int64_t p_length) const;
 	void _shape_run(ShapedTextDataAdvanced *p_sd, int64_t p_start, int64_t p_end, hb_script_t p_script, hb_direction_t p_direction, Array p_fonts, int64_t p_span, int64_t p_fb_index);
 	Glyph _shape_single_glyph(ShapedTextDataAdvanced *p_sd, char32_t p_char, hb_script_t p_script, hb_direction_t p_direction, const RID &p_font, int64_t p_font_size);
+
 	_FORCE_INLINE_ void _add_featuers(const Dictionary &p_source, Vector<hb_feature_t> &r_ftrs);
 
 	// HarfBuzz bitmap font interface.
@@ -686,6 +688,8 @@ 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 PackedInt32Array string_get_word_breaks(const String &p_string, const String &p_language = "") const override;
+
 	virtual String strip_diacritics(const String &p_string) const override;
 
 	virtual String string_to_upper(const String &p_string, const String &p_language = "") const override;
diff --git a/modules/text_server_fb/text_server_fb.cpp b/modules/text_server_fb/text_server_fb.cpp
index 1251aaf2b9..d84e9e581a 100644
--- a/modules/text_server_fb/text_server_fb.cpp
+++ b/modules/text_server_fb/text_server_fb.cpp
@@ -3079,7 +3079,7 @@ bool TextServerFallback::shaped_text_update_breaks(const RID &p_shaped) {
 		if (sd_glyphs[i].count > 0) {
 			char32_t c = sd->text[sd_glyphs[i].start - sd->start];
 			if (c_punct_size == 0) {
-				if (is_punct(c)) {
+				if (is_punct(c) && c != 0x005F) {
 					sd_glyphs[i].flags |= GRAPHEME_IS_PUNCTUATION;
 				}
 			} else {
@@ -3623,6 +3623,29 @@ String TextServerFallback::string_to_lower(const String &p_string, const String
 	return lower;
 }
 
+PackedInt32Array TextServerFallback::string_get_word_breaks(const String &p_string, const String &p_language) const {
+	PackedInt32Array ret;
+	for (int i = 0; i < p_string.length(); i++) {
+		char32_t c = p_string[i];
+		if (c == 0xfffc) {
+			continue;
+		}
+		if (is_punct(c) && c != 0x005F) {
+			ret.push_back(i);
+			continue;
+		}
+		if (is_underscore(c)) {
+			ret.push_back(i);
+			continue;
+		}
+		if (is_whitespace(c) || is_linebreak(c)) {
+			ret.push_back(i);
+			continue;
+		}
+	}
+	return ret;
+}
+
 TextServerFallback::TextServerFallback() {
 	_insert_feature_sets();
 };
diff --git a/modules/text_server_fb/text_server_fb.h b/modules/text_server_fb/text_server_fb.h
index d6f61e02f8..c837029623 100644
--- a/modules/text_server_fb/text_server_fb.h
+++ b/modules/text_server_fb/text_server_fb.h
@@ -573,6 +573,8 @@ public:
 	virtual double shaped_text_get_underline_position(const RID &p_shaped) const override;
 	virtual double shaped_text_get_underline_thickness(const RID &p_shaped) const override;
 
+	virtual PackedInt32Array string_get_word_breaks(const String &p_string, const String &p_language = "") const override;
+
 	virtual String string_to_upper(const String &p_string, const String &p_language = "") const override;
 	virtual String string_to_lower(const String &p_string, const String &p_language = "") const override;