perf(scanner): skip empty lines, grow stringbuilder by default

this change decreases the total runtime for around 1.2k lines by 0.2-0.5ms

Author: xnacly

scanner/scanner.go

@@ -183,8 +183,12 @@ func (s *Scanner) Parse() {
 			// PERF: option no3:
 			// 1.0ms, 1.2k lines, 4.5k token
 			var res strings.Builder
+
+			// PERF: growing the string build to the len of the current line - the current pos on the line keeps go from having to grow the builder itself
+			res.Grow(len(s.curLine) - int(s.linePos))
 		out:
 			for {
+				// PERF: using a switch instead of a hashmap decreases the runtime by around 50% for large files
 				switch s.curChar {
 				case '\n', '!', '#', '_', '*', '-', '[', ']', '(', ')', '`', '>':
 					break out
@@ -194,7 +198,10 @@ func (s *Scanner) Parse() {
 				s.advance()
 			}
 
-			s.addToken(TEXT, res.String())
+			// skip empty texts
+			if res.Len() != 0 {
+				s.addToken(TEXT, res.String())
+			}
 
 			// PERF: performing this here instead of in the next loop interation decreases execution time by around 0.1ms
 			if s.curChar == '\n' {