Use string_view for parameters to fromWide and toWide and for return …

…from RectangularSelection_Row::text. Make sure fromWide and toWide don't split long text between bytes of a multi-byte character.
Coises · Oct 16, 2023 · b3491c1 · b3491c1
1 parent eeb8c43
commit b3491c1
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 14 deletions.
diff --git a/src/ColumnsPlusPlus.h b/src/ColumnsPlusPlus.h
@@ -52,18 +52,20 @@ template< class T, class S > constexpr const T clamp_cast(const S& v) {
 }
 
 
-inline std::string fromWide(const std::wstring& s, unsigned int codepage) {
+inline std::string fromWide(std::wstring_view s, unsigned int codepage) {
     std::string r;
     size_t inputLength = s.length();
     if (!inputLength) return r;
     constexpr unsigned int safeSize = std::numeric_limits<int>::max() / 8;
     size_t workingPoint = 0;
     while (inputLength - workingPoint > safeSize) {
-        int segmentLength = WideCharToMultiByte(codepage, 0, s.data() + workingPoint, safeSize, 0, 0, 0, 0);
+        int ss = safeSize;
+        if (s[ss - 1] >= 0xD800 && s[ss - 1] <= 0xDBFF) --ss;  // proposed block ends with high surrogate: leave it for the next block
+        int segmentLength = WideCharToMultiByte(codepage, 0, s.data() + workingPoint, ss, 0, 0, 0, 0);
         size_t outputPoint = r.length();
         r.resize(outputPoint + segmentLength);
-        WideCharToMultiByte(codepage, 0, s.data() + workingPoint, safeSize, r.data() + outputPoint, segmentLength, 0, 0);
-        workingPoint += safeSize;
+        WideCharToMultiByte(codepage, 0, s.data() + workingPoint, ss, r.data() + outputPoint, segmentLength, 0, 0);
+        workingPoint += ss;
     }
     int segmentLength = WideCharToMultiByte(codepage, 0, s.data() + workingPoint, static_cast<int>(inputLength - workingPoint), 0, 0, 0, 0);
     size_t outputPoint = r.length();
@@ -72,18 +74,20 @@ inline std::string fromWide(const std::wstring& s, unsigned int codepage) {
     return r;
 }
 
-inline std::wstring toWide(const std::string& s, unsigned int codepage) {
+inline std::wstring toWide(std::string_view s, unsigned int codepage) {
     std::wstring r;
     size_t inputLength = s.length();
     if (!inputLength) return r;
     constexpr unsigned int safeSize = std::numeric_limits<int>::max() / 2;
     size_t workingPoint = 0;
     while (inputLength - workingPoint > safeSize) {
-        int segmentLength = MultiByteToWideChar(codepage, 0, s.data() + workingPoint, safeSize, 0, 0);
+        int ss = safeSize;
+        if (codepage == CP_UTF8 && ((s[ss] & 0xC0) == 0x80)) while ((s[--ss] & 0xC0) == 0x80);  // find a first byte to start the next block
+        int segmentLength = MultiByteToWideChar(codepage, 0, s.data() + workingPoint, ss, 0, 0);
         size_t outputPoint = r.length();
         r.resize(outputPoint + segmentLength);
-        MultiByteToWideChar(codepage, 0, s.data() + workingPoint, safeSize, r.data() + outputPoint, segmentLength);
-        workingPoint += safeSize;
+        MultiByteToWideChar(codepage, 0, s.data() + workingPoint, ss, r.data() + outputPoint, segmentLength);
+        workingPoint += ss;
     }
     int segmentLength = MultiByteToWideChar(codepage, 0, s.data() + workingPoint, static_cast<int>(inputLength - workingPoint), 0, 0);
     size_t outputPoint = r.length();
@@ -645,7 +649,7 @@ class RectangularSelection_Row {
     Scintilla::Position vsMax      () const { return std::max(_vsAnchor, _vsCaret); }
     Scintilla::Line     line       () const { return _line; }
     Scintilla::Position endOfLine  () const { return _endOfLine; }
-    std::string         text       () const { return _offset ? _text.substr(_offset) : _text ; }
+    std::string_view    text       () const { return std::string_view(_text).substr(_offset); }
     bool                isEndOfLine() const { return cpMax() == _endOfLine; }
     RectangularSelection_Cell_Iterator begin() const;
     RectangularSelection_Cell_Iterator end() const;

diff --git a/src/Configuration.cpp b/src/Configuration.cpp
@@ -106,7 +106,7 @@ void ColumnsPlusPlusData::loadConfiguration() {
         else if (std::regex_match(line, match, extensionsHeader  )) readingSection = sectionExtensions;
         else if (std::regex_match(line, match, profileHeader     )) {
             readingSection = sectionProfile;
-            profileName = toWide(match[1], CP_UTF8);
+            profileName = toWide(match.str(1), CP_UTF8);
         }
         else if (std::regex_match(line, match, dataLine)) {
             if (readingSection == sectionLastSettings) {
@@ -272,8 +272,8 @@ void ColumnsPlusPlusData::loadConfiguration() {
                 }
             }
             else if (readingSection == sectionExtensions) {
-                std::wstring extension = toWide(match[1], CP_UTF8);
-                std::wstring profile   = toWide(match[2], CP_UTF8);
+                std::wstring extension = toWide(match.str(1), CP_UTF8);
+                std::wstring profile   = toWide(match.str(2), CP_UTF8);
                 if      (extension == L"new")     extensionToProfile[L"" ] = profile;
                 else if (extension == L"none")    extensionToProfile[L"."] = profile;
                 else if (extension == L"default") extensionToProfile[L"*"] = profile;

diff --git a/src/Numeric.cpp b/src/Numeric.cpp
@@ -644,12 +644,14 @@ void ColumnsPlusPlusData::calculate() {
             if (!settings.elasticEnabled || !calc.tabbed) s += std::string(maxLeft + maxRight - s.length(), ' ');
             if (calc.left) {
                 s += calc.tabbed ? '\t' : ' ';
-                row.replace(s + row.text());
+                s += row.text();
+                row.replace(s);
             }
             else {
                 if (calc.tabbed) s = row.text().length() > 0 && row.text()[row.text().length() - 1] == '\t' ? s + '\t' : '\t' + s;
                 else s = ' ' + s;
-                row.replace(row.text() + s);
+                s.insert(0, row.text());
+                row.replace(s);
             }
         }
         ++resultsIndex;