Replace utf8 iterators with utf8_decode (fix #3260)

This should fix some problems decoding invalid UTF-8 strings.

laf

@@ -1 +1 @@
-Subproject commit 4f4693a8e63b7bc63e3c0825112f7f3f190ebee9
+Subproject commit df53f4ac0cecada789bf84b4283947d2591833bd

src/app/file/split_filename.cpp

@@ -1,4 +1,5 @@
 // Aseprite
+// Copyright (C) 2022  Igara Studio S.A.
 // Copyright (C) 2001-2017  David Capello
 //
 // This program is distributed under the terms of
@@ -9,9 +10,7 @@
 #endif
 
 #include "app/file/split_filename.h"
-#include "base/convert_to.h"
 #include "base/fs.h"
-#include "base/string.h"
 
 #include <cstring>
 
@@ -20,41 +19,44 @@ namespace app {
 // Splits a file-name like "my_ani0000.pcx" to "my_ani" and ".pcx",
 // returning the number of the center; returns "-1" if the function
 // can't split anything
-int split_filename(const std::string& filename, std::string& left, std::string& right, int& width)
+int split_filename(const std::string& filename,
+                   std::string& left,
+                   std::string& right,
+                   int& width)
 {
   left = base::get_file_title_with_path(filename);
   right = base::get_file_extension(filename);
   if (!right.empty())
     right.insert(right.begin(), '.');
 
-  // Remove all trailing numbers in the "left" side, and pass they to "result_str".
+  // Remove all trailing numbers in the "left" side.
   std::string result_str;
   width = 0;
-  for (;;) {
+  int num = -1;
-    // Get the last UTF-8 character (as we don't have a
+  int order = 1;
-    // reverse_iterator, we iterate from the beginning)
-    int chr = 0;
-    base::utf8_const_iterator begin(left.begin()), end(left.end());
-    base::utf8_const_iterator it(begin), prev(begin);
-    for (; it != end; prev=it, ++it)
-      chr = *it;
 
-    if ((chr >= '0') && (chr <= '9')) {
+  auto it = left.rbegin();
-      result_str.insert(result_str.begin(), chr);
+  auto end = left.rend();
-      width++;
 
-      left.erase(prev - begin);
+  while (it != end) {
+    const int chr = *it;
+    if (chr >= '0' && chr <= '9') {
+      if (num < 0)
+        num = 0;
+
+      num += order*(chr-'0');
+      order *= 10;
+      ++width;
+      ++it;
     }
     else
       break;
   }
 
-  // Convert the "buf" to integer and return it.
+  if (width > 0)
-  if (!result_str.empty()) {
+    left.erase(left.end()-width, left.end());
-    return base::convert_to<int>(result_str);
+
-  }
+  return num;
-  else
-    return -1;
 }
 
 } // namespace app

src/app/file/split_filename.h

@@ -1,4 +1,5 @@
 // Aseprite
+// Copyright (C) 2022  Igara Studio S.A.
 // Copyright (C) 2001-2017  David Capello
 //
 // This program is distributed under the terms of
@@ -12,7 +13,10 @@
 
 namespace app {
 
-  int split_filename(const std::string& filename, std::string& left, std::string& right, int& width);
+  int split_filename(const std::string& filename,
+                     std::string& left,
+                     std::string& right,
+                     int& width);
 
 } // namespace app
 

src/app/file/split_filename_tests.cpp

@@ -1,4 +1,5 @@
 // Aseprite
+// Copyright (C) 2022  Igara Studio S.A.
 // Copyright (C) 2001-2016  David Capello
 //
 // This program is distributed under the terms of
@@ -17,11 +18,21 @@ TEST(SplitFilename, Common)
   std::string left, right;
   int width;
 
+  EXPECT_EQ(-1, split_filename("sprite.png", left, right, width));
+  EXPECT_EQ("sprite", left);
+  EXPECT_EQ(".png", right);
+  EXPECT_EQ(0, width);
+
   EXPECT_EQ(1, split_filename("C:\\test\\a1.png", left, right, width));
   EXPECT_EQ("C:\\test\\a", left);
   EXPECT_EQ(".png", right);
   EXPECT_EQ(1, width);
 
+  EXPECT_EQ(2001, split_filename("/hi/bye2001.png", left, right, width));
+  EXPECT_EQ("/hi/bye", left);
+  EXPECT_EQ(".png", right);
+  EXPECT_EQ(4, width);
+
   EXPECT_EQ(1, split_filename("C:/test/a1.png", left, right, width));
   EXPECT_EQ("C:/test/a", left);
   EXPECT_EQ(".png", right);

src/app/ui/skin/skin_theme.cpp

@@ -29,6 +29,7 @@
 #include "base/fs.h"
 #include "base/log.h"
 #include "base/string.h"
+#include "base/utf8_decode.h"
 #include "gfx/border.h"
 #include "gfx/point.h"
 #include "gfx/rect.h"
@@ -1131,14 +1132,13 @@ void SkinTheme::drawEntryText(ui::Graphics* g, ui::Entry* widget)
   int scroll = delegate.index();
 
   const std::string& textString = widget->text();
-  base::utf8_const_iterator utf8_it((textString.begin()));
+  base::utf8_decode dec(textString);
-  int textlen = base::utf8_length(textString);
+  auto pos = dec.pos();
-  scroll = std::min(scroll, textlen);
+  for (int i=0; i<scroll && dec.next(); ++i)
-  if (scroll)
+    pos = dec.pos();
-    utf8_it += scroll;
 
-  g->drawText(utf8_it,
+  // TODO use a string_view()
-              base::utf8_const_iterator(textString.end()),
+  g->drawText(std::string(pos, textString.end()),
               colors.text(), ColorNone,
               bounds.origin(), &delegate);
 

src/app/util/freetype_utils.cpp

@@ -1,4 +1,5 @@
 // Aseprite
+// Copyright (C) 2022  Igara Studio S.A.
 // Copyright (C) 2001-2018  David Capello
 //
 // This program is distributed under the terms of
@@ -47,10 +48,8 @@ doc::Image* render_text(const std::string& fontfile, int fontsize,
       image.reset(doc::Image::create(doc::IMAGE_RGB, bounds.w, bounds.h));
       doc::clear_image(image.get(), 0);
 
-      ft::ForEachGlyph<ft::Face> feg(face);
+      ft::ForEachGlyph<ft::Face> feg(face, text);
-      if (feg.initialize(base::utf8_const_iterator(text.begin()),
+      while (feg.next()) {
-                         base::utf8_const_iterator(text.end()))) {
-        do {
         auto glyph = feg.glyph();
         if (!glyph)
           continue;
@@ -92,7 +91,6 @@ doc::Image* render_text(const std::string& fontfile, int fontsize,
             }
           }
         }
-        } while (feg.nextChar());
       }
     }
     else {

src/ui/entry.cpp

@@ -922,9 +922,7 @@ void Entry::recalcCharBoxes(const std::string& text)
 {
   int lastTextIndex = int(text.size());
   CalcBoxesTextDelegate delegate(lastTextIndex);
-  os::draw_text(nullptr, font(),
+  os::draw_text(nullptr, font(), text,
-                 base::utf8_const_iterator(text.begin()),
-                 base::utf8_const_iterator(text.end()),
                  gfx::ColorNone, gfx::ColorNone, 0, 0, &delegate);
   m_boxes = delegate.boxes();
 

src/ui/graphics.cpp

@@ -362,8 +362,7 @@ void Graphics::setFont(const os::FontRef& font)
   m_font = font;
 }
 
-void Graphics::drawText(base::utf8_const_iterator it,
+void Graphics::drawText(const std::string& str,
-                        const base::utf8_const_iterator& end,
                         gfx::Color fg, gfx::Color bg,
                         const gfx::Point& origPt,
                         os::DrawTextDelegate* delegate)
@@ -372,18 +371,11 @@ void Graphics::drawText(base::utf8_const_iterator it,
 
   os::SurfaceLock lock(m_surface.get());
   gfx::Rect textBounds =
-    os::draw_text(m_surface.get(), m_font.get(), it, end, fg, bg, pt.x, pt.y, delegate);
+    os::draw_text(m_surface.get(), m_font.get(), str, fg, bg, pt.x, pt.y, delegate);
 
   dirty(gfx::Rect(pt.x, pt.y, textBounds.w, textBounds.h));
 }
 
-void Graphics::drawText(const std::string& str, gfx::Color fg, gfx::Color bg, const gfx::Point& pt)
-{
-  drawText(base::utf8_const_iterator(str.begin()),
-           base::utf8_const_iterator(str.end()),
-           fg, bg, pt, nullptr);
-}
-
 namespace {
 
 class DrawUITextDelegate : public os::DrawTextDelegate {
@@ -457,9 +449,7 @@ void Graphics::drawUIText(const std::string& str, gfx::Color fg, gfx::Color bg,
   int y = m_dy+pt.y;
 
   DrawUITextDelegate delegate(m_surface.get(), m_font.get(), mnemonic);
-  os::draw_text(m_surface.get(), m_font.get(),
+  os::draw_text(m_surface.get(), m_font.get(), str,
-                 base::utf8_const_iterator(str.begin()),
-                 base::utf8_const_iterator(str.end()),
                 fg, bg, x, y, &delegate);
 
   dirty(delegate.bounds());
@@ -482,9 +472,7 @@ gfx::Size Graphics::measureUIText(const std::string& str)
 int Graphics::measureUITextLength(const std::string& str, os::Font* font)
 {
   DrawUITextDelegate delegate(nullptr, font, 0);
-  os::draw_text(nullptr, font,
+  os::draw_text(nullptr, font, str,
-                 base::utf8_const_iterator(str.begin()),
-                 base::utf8_const_iterator(str.end()),
                 gfx::ColorNone, gfx::ColorNone, 0, 0,
                 &delegate);
   return delegate.bounds().w;

src/ui/graphics.h

@@ -117,11 +117,10 @@ namespace ui {
     os::Font* font() { return m_font.get(); }
     void setFont(const os::FontRef& font);
 
-    void drawText(base::utf8_const_iterator it,
+    void drawText(const std::string& str,
-                  const base::utf8_const_iterator& end,
+                  gfx::Color fg, gfx::Color bg,
-                  gfx::Color fg, gfx::Color bg, const gfx::Point& pt,
+                  const gfx::Point& pt,
-                  os::DrawTextDelegate* delegate);
+                  os::DrawTextDelegate* delegate = nullptr);
-    void drawText(const std::string& str, gfx::Color fg, gfx::Color bg, const gfx::Point& pt);
     void drawUIText(const std::string& str, gfx::Color fg, gfx::Color bg, const gfx::Point& pt, const int mnemonic);
     void drawAlignedUIText(const std::string& str, gfx::Color fg, gfx::Color bg, const gfx::Rect& rc, const int align);
 

src/ui/widget.cpp

@@ -16,6 +16,7 @@
 #include "base/clamp.h"
 #include "base/memory.h"
 #include "base/string.h"
+#include "base/utf8_decode.h"
 #include "os/font.h"
 #include "os/surface.h"
 #include "os/system.h"
@@ -1400,19 +1401,18 @@ void Widget::processMnemonicFromText(int escapeChar)
   if (!m_text.empty())
     newText.reserve(m_text.size());
 
-  for (base::utf8_const_iterator
+  base::utf8_decode decode(m_text);
-         it(m_text.begin()),
+  while (int chr = decode.next()) {
-         end(m_text.end()); it != end; ++it) {
+    if (chr == escapeChar) {
-    if (*it == escapeChar) {
+      chr = decode.next();
-      ++it;
+      if (!chr) {
-      if (it == end) {
         break;    // Ill-formed string (it ends with escape character)
       }
-      else if (*it != escapeChar) {
+      else if (chr != escapeChar) {
-        setMnemonic(*it);
+        setMnemonic(chr);
       }
     }
-    newText.push_back(*it);
+    newText.push_back(chr);
   }
 
   setText(base::to_utf8(newText));