Merge pull request #44360 from bruvzg/ctl_punct_word_break

Add word breaks on punctuation characters.

doc/classes/TextServer.xml

@@ -1167,6 +1167,9 @@
 		<constant name="GRAPHEME_IS_ELONGATION" value="128" enum="GraphemeFlag">
 			Grapheme is kashida.
 		</constant>
+		<constant name="GRAPHEME_IS_PUNCTUATION" value="256" enum="GraphemeFlag">
+			Grapheme is punctuation character.
+		</constant>
 		<constant name="HINTING_NONE" value="0" enum="Hinting">
 			Disables font hinting (smoother but less crisp).
 		</constant>

modules/text_server_adv/text_server_adv.cpp

@@ -1645,13 +1645,18 @@ bool TextServerAdvanced::shaped_text_update_breaks(RID p_shaped) {
 			if (c == 0x0009 || c == 0x000b) {
 				sd_glyphs[i].flags |= GRAPHEME_IS_TAB;
 			}
+			if (is_whitespace(c)) {
+				sd_glyphs[i].flags |= GRAPHEME_IS_SPACE;
+			}
+			if (u_ispunct(c)) {
+				sd_glyphs[i].flags |= GRAPHEME_IS_PUNCTUATION;
+			}
 			if (breaks.has(sd->glyphs[i].start)) {
 				if (breaks[sd->glyphs[i].start]) {
 					sd_glyphs[i].flags |= GRAPHEME_IS_BREAK_HARD;
 				} else {
 					if (is_whitespace(c)) {
 						sd_glyphs[i].flags |= GRAPHEME_IS_BREAK_SOFT;
-						sd_glyphs[i].flags |= GRAPHEME_IS_SPACE;
 					} else {
 						TextServer::Glyph gl;
 						gl.start = sd_glyphs[i].start;
@@ -1766,6 +1771,10 @@ bool TextServerAdvanced::shaped_text_update_justification_ops(RID p_shaped) {
 		shaped_text_update_breaks(p_shaped);
 	}
 
+	if (sd->justification_ops_valid) {
+		return true; // Noting to do.
+	}
+
 	const UChar *data = sd->utf16.ptr();
 	int32_t data_size = sd->utf16.length();
 
@@ -1796,9 +1805,9 @@ bool TextServerAdvanced::shaped_text_update_justification_ops(RID p_shaped) {
 			if (ubrk_getRuleStatus(bi) != UBRK_WORD_NONE) {
 				int i = _convert_pos(sd, ubrk_current(bi));
 				jstops[i + sd->start] = false;
-				int ks = _generate_kashida_justification_opportunies(sd->text, limit, i) + sd->start;
+				int ks = _generate_kashida_justification_opportunies(sd->text, limit, i);
 				if (ks != -1) {
-					jstops[ks] = true;
+					jstops[ks + sd->start] = true;
 				}
 				limit = i;
 			}

modules/text_server_fb/text_server_fb.cpp

@@ -45,6 +45,10 @@ _FORCE_INLINE_ bool is_linebreak(char32_t p_char) {
 	return (p_char >= 0x000a && p_char <= 0x000d) || (p_char == 0x0085) || (p_char == 0x2028) || (p_char == 0x2029);
 }
 
+_FORCE_INLINE_ bool is_punct(char32_t p_char) {
+	return (p_char >= 0x0020 && p_char <= 0x002F) || (p_char >= 0x003A && p_char <= 0x0040) || (p_char >= 0x005B && p_char <= 0x0060) || (p_char >= 0x007B && p_char <= 0x007E) || (p_char >= 0x2000 && p_char <= 0x206F) || (p_char >= 0x3000 && p_char <= 0x303F);
+}
+
 /*************************************************************************/
 
 String TextServerFallback::interface_name = "Fallback";
@@ -1029,6 +1033,9 @@ bool TextServerFallback::shaped_text_update_breaks(RID p_shaped) {
 	for (int i = 0; i < sd_size; i++) {
 		if (sd->glyphs[i].count > 0) {
 			char32_t c = sd->text[sd->glyphs[i].start];
+			if (is_punct(c)) {
+				sd->glyphs.write[i].flags |= GRAPHEME_IS_PUNCTUATION;
+			}
 			if (is_whitespace(c) && !is_linebreak(c)) {
 				sd->glyphs.write[i].flags |= GRAPHEME_IS_SPACE;
 				sd->glyphs.write[i].flags |= GRAPHEME_IS_BREAK_SOFT;

servers/text_server.cpp

@@ -376,6 +376,7 @@ void TextServer::_bind_methods() {
 	BIND_ENUM_CONSTANT(GRAPHEME_IS_BREAK_SOFT);
 	BIND_ENUM_CONSTANT(GRAPHEME_IS_TAB);
 	BIND_ENUM_CONSTANT(GRAPHEME_IS_ELONGATION);
+	BIND_ENUM_CONSTANT(GRAPHEME_IS_PUNCTUATION);
 
 	/* Hinting */
 	BIND_ENUM_CONSTANT(HINTING_NONE);
@@ -679,7 +680,7 @@ Vector<Vector2i> TextServer::shaped_text_get_line_breaks(RID p_shaped, float p_w
 Vector<Vector2i> TextServer::shaped_text_get_word_breaks(RID p_shaped) const {
 	Vector<Vector2i> words;
 
-	const_cast<TextServer *>(this)->shaped_text_update_breaks(p_shaped);
+	const_cast<TextServer *>(this)->shaped_text_update_justification_ops(p_shaped);
 	const Vector<Glyph> &logical = const_cast<TextServer *>(this)->shaped_text_sort_logical(p_shaped);
 	const Vector2i &range = shaped_text_get_range(p_shaped);
 
@@ -690,8 +691,8 @@ Vector<Vector2i> TextServer::shaped_text_get_word_breaks(RID p_shaped) const {
 
 	for (int i = 0; i < l_size; i++) {
 		if (l_gl[i].count > 0) {
-			if ((l_gl[i].flags & GRAPHEME_IS_SPACE) == GRAPHEME_IS_SPACE) {
-				words.push_back(Vector2i(word_start, l_gl[i].end - 1));
+			if (((l_gl[i].flags & GRAPHEME_IS_SPACE) == GRAPHEME_IS_SPACE) || ((l_gl[i].flags & GRAPHEME_IS_PUNCTUATION) == GRAPHEME_IS_PUNCTUATION)) {
+				words.push_back(Vector2i(word_start, l_gl[i].start));
 				word_start = l_gl[i].end;
 			}
 		}

servers/text_server.h

@@ -74,11 +74,12 @@ public:
 		GRAPHEME_IS_VALID = 1 << 0, // Glyph is valid.
 		GRAPHEME_IS_RTL = 1 << 1, // Glyph is right-to-left.
 		GRAPHEME_IS_VIRTUAL = 1 << 2, // Glyph is not part of source string (added by fit_to_width function, do not affect caret movement).
-		GRAPHEME_IS_SPACE = 1 << 3, // Is whitespace (for justification).
-		GRAPHEME_IS_BREAK_HARD = 1 << 4, // Is line break (mandatory break, e.g "\n")
-		GRAPHEME_IS_BREAK_SOFT = 1 << 5, // Is line break (optional break, e.g space)
-		GRAPHEME_IS_TAB = 1 << 6, // Is tab or vertical tab
-		GRAPHEME_IS_ELONGATION = 1 << 7 // Elongation (e.g kashida), glyph can be duplicated or truncated to fit line to width.
+		GRAPHEME_IS_SPACE = 1 << 3, // Is whitespace (for justification and word breaks).
+		GRAPHEME_IS_BREAK_HARD = 1 << 4, // Is line break (mandatory break, e.g. "\n").
+		GRAPHEME_IS_BREAK_SOFT = 1 << 5, // Is line break (optional break, e.g. space).
+		GRAPHEME_IS_TAB = 1 << 6, // Is tab or vertical tab.
+		GRAPHEME_IS_ELONGATION = 1 << 7, // Elongation (e.g. kashida), glyph can be duplicated or truncated to fit line to width.
+		GRAPHEME_IS_PUNCTUATION = 1 << 8 // Punctuation (can be used as word break, but not line break or justifiction).
 	};
 
 	enum Hinting {
@@ -104,7 +105,7 @@ public:
 
 		uint8_t count = 0; // Number of glyphs in the grapheme, set in the first glyph only.
 		uint8_t repeat = 1; // Draw multiple times in the row.
-		uint8_t flags = 0; // Grapheme flags (valid, rtl, virtual), set in the first glyph only.
+		uint16_t flags = 0; // Grapheme flags (valid, rtl, virtual), set in the first glyph only.
 
 		float x_off = 0.f; // Offset from the origin of the glyph on baseline.
 		float y_off = 0.f;