string_view version of utf8_display_split to optimize mapgen with way…

… fewer allocs
CleverRaven · Dec 24, 2023 · f40d35d · f40d35d
1 parent d0c725e
commit f40d35d
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 13 deletions.
diff --git a/src/catacharset.cpp b/src/catacharset.cpp
@@ -415,21 +415,31 @@ std::u32string utf8_to_utf32( const std::string_view str )
 std::vector<std::string> utf8_display_split( const std::string &s )
 {
     std::vector<std::string> result;
-    std::string current_glyph;
+    std::vector<std::string_view> parts;
+    utf8_display_split_into( s, parts );
+    result.reserve( parts.size() );
+    for( std::string_view part : parts ) {
+        result.emplace_back( part );
+    }
+    return result;
+}
+
+void utf8_display_split_into( const std::string &s, std::vector<std::string_view> &result )
+{
     const char *pos = s.c_str();
+    const char *glyph_begin = pos;
+    const char *glyph_end = pos;
     int len = s.length();
     while( len > 0 ) {
-        const char *old_pos = pos;
         const uint32_t ch = UTF8_getch( &pos, &len );
         const int width = mk_wcwidth( ch );
-        if( width > 0 && !current_glyph.empty() ) {
-            result.push_back( current_glyph );
-            current_glyph.clear();
+        if( width > 0 && glyph_begin != glyph_end ) {
+            result.push_back( std::string_view( glyph_begin, std::distance( glyph_begin, glyph_end ) ) );
+            glyph_begin = glyph_end;
         }
-        current_glyph += std::string( old_pos, pos );
+        glyph_end = pos;
     }
-    result.push_back( current_glyph );
-    return result;
+    result.push_back( std::string_view( glyph_begin, std::distance( glyph_begin, glyph_end ) ) );
 }
 
 int center_text_pos( const char *text, int start_pos, int end_pos )

diff --git a/src/catacharset.h b/src/catacharset.h
@@ -59,6 +59,7 @@ std::u32string utf8_to_utf32( std::string_view str );
 // Split the given string into displayed characters.  Each element of the returned vector
 // contains one 'regular' codepoint and all subsequent combining characters.
 std::vector<std::string> utf8_display_split( const std::string & );
+void utf8_display_split_into( const std::string &, std::vector<std::string_view> & );
 
 /**
  * UTF8-Wrapper over std::string.

diff --git a/src/mapgen.cpp b/src/mapgen.cpp
@@ -4491,10 +4491,10 @@ bool mapgen_function_json_base::setup_common( const JsonObject &jo )
         }
         for( int c = m_offset.y; c < expected_dim.y; c++ ) {
             const std::string row = parray.get_string( c );
-            std::vector<map_key> row_keys;
-            for( const std::string &key : utf8_display_split( row ) ) {
-                row_keys.emplace_back( key );
-            }
+            static std::vector<std::string_view> row_keys;
+            row_keys.clear();
+            row_keys.reserve( total_size.x );
+            utf8_display_split_into( row, row_keys );
             if( row_keys.size() < static_cast<size_t>( expected_dim.x ) ) {
                 parray.throw_error(
                     string_format( "  format: row %d must have at least %d columns, not %d",
@@ -4507,7 +4507,7 @@ bool mapgen_function_json_base::setup_common( const JsonObject &jo )
             }
             for( int i = m_offset.x; i < expected_dim.x; i++ ) {
                 const point p = point( i, c ) - m_offset;
-                const map_key key = row_keys[i];
+                const map_key key{ std::string( row_keys[i] ) };
                 const auto iter_ter = keys_with_terrain.find( key );
                 const auto fpi = format_placings.find( key );