From: Mikko Rasa <tdb@tdb.fi>
Date: Tue, 17 Sep 2019 16:06:34 +0000 (+0300)
Subject: Properly handle multibyte UTF-8 in Entry and Text
X-Git-Url: http://git.tdb.fi/?p=libs%2Fgltk.git;a=commitdiff_plain;h=2aa99730d8c3106deeac1186e057055604835752

Properly handle multibyte UTF-8 in Entry and Text
---

diff --git a/source/entry.cpp b/source/entry.cpp
index 8b638de..6554f7c 100644
--- a/source/entry.cpp
+++ b/source/entry.cpp
@@ -269,14 +269,20 @@ bool Entry::key_press(unsigned key, unsigned mod)
 		if(selection_active)
 			erase_selection(true);
 		else if(edit_pos>0)
-			erase(edit_pos-1, 1);
+		{
+			unsigned start_pos = text.move_offset(edit_pos, -1);
+			erase(start_pos, edit_pos-start_pos);
+		}
 	}
 	else if(key==Input::KEY_DELETE)
 	{
 		if(selection_active)
 			erase_selection(true);
 		else
-			erase(edit_pos, 1);
+		{
+			unsigned end_pos = text.move_offset(edit_pos, 1);
+			erase(edit_pos, end_pos-edit_pos);
+		}
 	}
 	else if(key==Input::KEY_ENTER && multiline)
 		insert(edit_pos, "\n");
@@ -378,15 +384,9 @@ void Entry::on_style_change()
 void Entry::move_edit_position(Navigation nav, bool select)
 {
 	if(nav==NAV_LEFT)
-	{
-		if(edit_pos>0)
-			set_edit_position(edit_pos-1, select);
-	}
+		set_edit_position(text.move_offset(edit_pos, -1), select);
 	else if(nav==NAV_RIGHT)
-	{
-		if(edit_pos<text.size())
-			set_edit_position(edit_pos+1, select);
-	}
+		set_edit_position(text.move_offset(edit_pos, 1), select);
 	else if(nav==NAV_DOWN)
 	{
 		unsigned row, col;
@@ -399,6 +399,8 @@ void Entry::move_edit_position(Navigation nav, bool select)
 		text.offset_to_coords(edit_pos, row, col);
 		set_edit_position((row>0 ? text.coords_to_offset(row-1, col) : 0), select);
 	}
+	else
+		throw invalid_argument("Entry::move_edit_position");
 }
 
 void Entry::set_edit_position(unsigned ep, bool select)
diff --git a/source/text.cpp b/source/text.cpp
index 45b6567..a3ed762 100644
--- a/source/text.cpp
+++ b/source/text.cpp
@@ -42,7 +42,7 @@ void Text::set_style(const Style *s)
 		const GL::Font &font = style->get_font();
 		float font_size = style->get_font_size();
 		for(vector<Line>::iterator i=lines.begin(); i!=lines.end(); ++i)
-			i->width = static_cast<unsigned>(font.get_string_width(text.substr(i->start, i->length))*font_size);
+			i->width = static_cast<unsigned>(font.get_string_width(text.substr(i->start, i->bytes))*font_size);
 	}
 }
 
@@ -81,16 +81,19 @@ void Text::set(const string &t)
 
 void Text::erase(unsigned pos, unsigned len)
 {
+	check_alignment(pos);
+	check_alignment(pos+len);
 	text.erase(pos, len);
 
 	vector<Line>::iterator i;
-	for(i=lines.begin(); (i!=lines.end() && i->start+i->length<pos); ++i) ;
+	for(i=lines.begin(); (i!=lines.end() && i->start+i->bytes<pos); ++i) ;
 
-	if(pos+len>i->start+i->length)
+	if(pos+len>i->start+i->bytes)
 		find_lines();
 	else
 	{
-		i->length -= len;
+		i->bytes -= len;
+		i->length = count_characters(i->start, i->bytes);
 
 		for(++i; i!=lines.end(); ++i)
 			i->start -= len;
@@ -99,6 +102,7 @@ void Text::erase(unsigned pos, unsigned len)
 
 void Text::insert(unsigned pos, const string &s)
 {
+	check_alignment(pos);
 	text.insert(pos, s);
 
 	if(s.find('\n')!=string::npos)
@@ -106,9 +110,10 @@ void Text::insert(unsigned pos, const string &s)
 	else
 	{
 		vector<Line>::iterator i;
-		for(i=lines.begin(); (i!=lines.end() && i->start+i->length<pos); ++i) ;
+		for(i=lines.begin(); (i!=lines.end() && i->start+i->bytes<pos); ++i) ;
 
-		i->length += s.size();
+		i->bytes += s.size();
+		i->length = count_characters(i->start, i->bytes);
 
 		for(++i; i!=lines.end(); ++i)
 			i->start += s.size();
@@ -139,6 +144,32 @@ unsigned Text::get_line_length(unsigned i) const
 	return lines[i].length;
 }
 
+unsigned Text::move_offset(unsigned offs, int change) const
+{
+	check_alignment(offs);
+	if(!change)
+		return offs;
+
+	StringCodec::Utf8::Decoder dec(StringCodec::IGNORE_ERRORS);
+	string::const_iterator i = text.begin()+offs;
+	if(change>0)
+	{
+		for(; change>0; --change)
+			dec.decode_char(text, i);
+	}
+	else
+	{
+		while(change<0 && i!=text.begin())
+		{
+			--i;
+			string::const_iterator j = i;
+			if(dec.decode_char(text, j)!=-1)
+				++change;
+		}
+	}
+	return i-text.begin();
+}
+
 void Text::offset_to_coords(unsigned offs, unsigned &row, unsigned &col) const
 {
 	if(lines.empty())
@@ -149,10 +180,13 @@ void Text::offset_to_coords(unsigned offs, unsigned &row, unsigned &col) const
 	}
 
 	for(unsigned i=0; i<lines.size(); ++i)
-		if(offs>=lines[i].start && offs<=lines[i].start+lines[i].length)
+		if(offs>=lines[i].start && offs<=lines[i].start+lines[i].bytes)
 		{
 			row = i;
-			col = offs-lines[i].start;
+			if(lines[i].length==lines[i].bytes)
+				col = offs-lines[i].start;
+			else
+				col = count_characters(lines[i].start, offs-lines[i].start);
 			return;
 		}
 }
@@ -161,8 +195,20 @@ unsigned Text::coords_to_offset(unsigned row, unsigned col) const
 {
 	if(row>=lines.size())
 		return text.size();
+	const Line &line = lines[row];
+	if(col>line.length)
+		col = line.length;
 
-	return lines[row].start+min(col, lines[row].length);
+	if(line.length==line.bytes)
+		return line.start+col;
+	else
+	{
+		StringCodec::Utf8::Decoder dec;
+		string::const_iterator i = text.begin()+line.start;
+		for(col=min(col, line.length); col; --col)
+			dec.decode_char(text, i);
+		return i-text.begin();
+	}
 }
 
 Geometry Text::coords_to_geometry(const Part &part, const Geometry &parent, unsigned first_row, unsigned row, unsigned col) const
@@ -219,11 +265,12 @@ void Text::find_lines()
 
 		Line line;
 		line.start = start;
-		line.length = (newline==string::npos ? text.size() : newline)-start;
+		line.bytes = (newline==string::npos ? text.size() : newline)-start;
+		line.length = count_characters(line.start, line.bytes);
 		line.width = line.length;
 		if(style)
 		{
-			string str = text.substr(line.start, line.length);
+			string str = text.substr(line.start, line.bytes);
 			line.width = static_cast<unsigned>(style->get_font().get_string_width(str)*font_size);
 		}
 		lines.push_back(line);
@@ -234,6 +281,24 @@ void Text::find_lines()
 	}
 }
 
+unsigned Text::count_characters(unsigned start, unsigned bytes) const
+{
+	StringCodec::Utf8::Decoder dec;
+	string::const_iterator i = text.begin()+start;
+	string::const_iterator end = i+bytes;
+	unsigned count = 0;
+	for(; i<end; dec.decode_char(text, i))
+		++count;
+	return count;
+}
+
+void Text::check_alignment(unsigned offs) const
+{
+	StringCodec::Utf8::Decoder dec;
+	string::const_iterator i = text.begin()+offs;
+	dec.decode_char(text, i);
+}
+
 template<typename T>
 void Text::process_lines(const Part &part, const Geometry &parent, unsigned first_row, void (Text::*func)(unsigned, const Geometry &, T &) const, T &data) const
 {
@@ -271,14 +336,24 @@ void Text::build_line(unsigned i, const Geometry &rgeom, RenderData &data) const
 	data.bld->matrix() *= GL::Matrix::translation(rgeom.x, rgeom.y, 0);
 	data.bld->matrix() *= GL::Matrix::scaling(style->get_font_size());
 
-	style->get_font().build_string(text.substr(line.start, line.length), *data.bld);
+	style->get_font().build_string(text.substr(line.start, line.bytes), *data.bld);
 }
 
 void Text::coords_to_geom_line(unsigned i, const Geometry &rgeom, CoordsToGeomData &data) const
 {
 	if(i==data.row)
 	{
-		float w = style->get_font().get_string_width(text.substr(lines[i].start, data.col));
+		string::const_iterator begin = text.begin()+lines[i].start;
+		string::const_iterator j = begin;
+		if(lines[i].length==lines[i].bytes)
+			j += data.col;
+		else
+		{
+			StringCodec::Utf8::Decoder dec;
+			for(unsigned c=data.col; c; --c)
+				dec.decode_char(text, j);
+		}
+		float w = style->get_font().get_string_width(string(begin, j));
 		data.result = rgeom;
 		data.result.x += static_cast<unsigned>(w*style->get_font_size());
 	}
diff --git a/source/text.h b/source/text.h
index 243c3a6..6ddbb23 100644
--- a/source/text.h
+++ b/source/text.h
@@ -22,6 +22,7 @@ private:
 	struct Line
 	{
 		unsigned start;
+		unsigned bytes;
 		unsigned length;
 		unsigned width;
 	};
@@ -51,6 +52,7 @@ public:
 	unsigned get_n_lines() const { return lines.size(); }
 	unsigned get_visible_lines(const Part &, const Geometry &, unsigned *) const;
 	unsigned get_line_length(unsigned) const;
+	unsigned move_offset(unsigned, int) const;
 	void offset_to_coords(unsigned, unsigned &, unsigned &) const;
 	unsigned coords_to_offset(unsigned, unsigned) const;
 	Geometry coords_to_geometry(const Part &, const Geometry &, unsigned, unsigned, unsigned) const;
@@ -61,6 +63,8 @@ public:
 	Text &operator=(const std::string &);
 private:
 	void find_lines();
+	unsigned count_characters(unsigned, unsigned) const;
+	void check_alignment(unsigned) const;
 
 	template<typename T>
 	void process_lines(const Part &, const Geometry &, unsigned, void (Text::*)(unsigned, const Geometry &, T &) const, T &) const;