string performance optimisation

Author: sergeevik

src/main/java/com/dashjoin/jsonata/Functions.java

@@ -609,10 +609,11 @@ public static String leftPad(final String str, final int size, String padStr) {
         if (pads <= 0) {
             return str;
         }
-        String padding = "";
+        StringBuilder paddingSb = new StringBuilder();
         for (int i = 0; i < pads + 1; i++) {
-            padding += padStr;
+            paddingSb.append(padStr);
         }
+        String padding = paddingSb.toString();
         return substr(padding, 0, pads).concat(str);
     }
 
@@ -638,10 +639,11 @@ public static String rightPad(final String str, final int size, String padStr) {
         if (pads <= 0) {
             return str;
         }
-        String padding = "";
+        StringBuilder paddingSb = new StringBuilder();
         for (int i = 0; i < pads + 1; i++) {
-            padding += padStr;
+            paddingSb.append(padStr);
         }
+        String padding = paddingSb.toString();
         return str.concat(substr(padding, 0, pads));
     }
 
@@ -770,14 +772,21 @@ public static String join(List<String> strs, String separator) {
         return String.join(separator, strs);
     }
 
+    private static final Pattern DOLLAR_DOLLAR = Pattern.compile("\\$\\$");
+    private static final Pattern DOLLAR_WITHOUT_ESCAPE = Pattern.compile("([^\\\\]|^)\\$([^0-9^<])");
+    private static final Pattern DOLLAR_AT_END = Pattern.compile("\\$$");
     static String safeReplacement(String in) {
         // In JSONata and in Java the $ in the replacement test usually starts the insertion of a capturing group
         // In order to replace a simple $ in Java you have to escape the $ with "\$"
         // in JSONata you do this with a '$$'
-        // "\$" followed any character besides '<' and and digit into $ + this character  
-        return in.replaceAll("\\$\\$", "\\\\\\$")
-            .replaceAll("([^\\\\]|^)\\$([^0-9^<])", "$1\\\\\\$$2")
-            .replaceAll("\\$$", "\\\\\\$"); // allow $ at end
+        // "\$" followed any character besides '<' and and digit into $ + this character
+        if (!in.contains("$")) {
+            return in;
+        }
+        String result = DOLLAR_DOLLAR.matcher(in).replaceAll("\\\\\\$");
+        result = DOLLAR_WITHOUT_ESCAPE.matcher(result).replaceAll("$1\\\\\\$$2");
+        result = DOLLAR_AT_END.matcher(result).replaceAll("\\\\\\$");
+        return result;
     }
 
     /**
@@ -813,7 +822,7 @@ static String safeReplaceAll(String s, Pattern pattern, Object _replacement) {
                 if (!msg.contains("No group")) throw e;
 
                 // Adjust replacement to remove the non-existing group
-                String g = "" + msg.charAt(msg.length()-1);
+                String g = String.valueOf(msg.charAt(msg.length()-1));
 
                 replacement = replacement.replace("$"+g, "");
             }
@@ -865,7 +874,7 @@ static String safeReplaceAllFn(String s, Pattern pattern, Object fn) {
 
     /**
      * Safe replaceFirst
-     * 
+     *
      * @param s
      * @param pattern
      * @param replacement
@@ -887,7 +896,7 @@ static String safeReplaceFirst(String s, Pattern pattern, String replacement) {
                 if (!msg.contains("No group")) throw e;
 
                 // Adjust replacement to remove the non-existing group
-                String g = "" + msg.charAt(msg.length()-1);
+                String g = String.valueOf(msg.charAt(msg.length()-1));
 
                 replacement = replacement.replace("$"+g, "");
             }
@@ -909,10 +918,10 @@ public static String replace(String str, Object pattern, Object replacement, Int
               return safeReplaceAll(str, (Pattern)pattern, replacement);
           }
         } else {
-          
+
           if (limit<0)
             throw new JException("Fourth argument of replace function must evaluate to a positive number", 0);
-          
+
           for (int i=0; i<limit; i++)
             if (pattern instanceof String) {
                 str = str.replaceFirst((String)pattern, (String)replacement);
@@ -963,6 +972,12 @@ public static String base64decode(String str) {
         }
     }
 
+    private static final Pattern PLUS = Pattern.compile("\\+");
+    private static final Pattern PERCENT_21 = Pattern.compile("%21");
+    private static final Pattern PERCENT_27 = Pattern.compile("%27");
+    private static final Pattern PERCENT_28 = Pattern.compile("%28");
+    private static final Pattern PERCENT_29 = Pattern.compile("%29");
+    private static final Pattern PERCENT_7E = Pattern.compile("%7E");
     /**
      * Encode a string into a component for a url
      * @param {String} str - String to encode
@@ -975,14 +990,20 @@ public static String encodeUrlComponent(String str) {
         }
 
         Utils.checkUrl(str);
-        
-        return URLEncoder.encode(str, StandardCharsets.UTF_8)
-                            .replaceAll("\\+", "%20")
-                            .replaceAll("\\%21", "!")
-                            .replaceAll("\\%27", "'")
-                            .replaceAll("\\%28", "(")
-                            .replaceAll("\\%29", ")")
-                            .replaceAll("\\%7E", "~");
+
+        String encoded = URLEncoder.encode(str, StandardCharsets.UTF_8);
+
+        if (!encoded.contains("+") && !encoded.contains("%")) {
+            return encoded;
+        }
+
+        encoded = PLUS.matcher(encoded).replaceAll("%20");
+        encoded = PERCENT_21.matcher(encoded).replaceAll("!");
+        encoded = PERCENT_27.matcher(encoded).replaceAll("'");
+        encoded = PERCENT_28.matcher(encoded).replaceAll("(");
+        encoded = PERCENT_29.matcher(encoded).replaceAll(")");
+        encoded = PERCENT_7E.matcher(encoded).replaceAll("~");
+        return encoded;
     }
 
     /**
@@ -997,7 +1018,7 @@ public static String encodeUrl(String str) {
         }
 
         Utils.checkUrl(str);
-        
+
         try {
           // only encode query part: https://docs.jsonata.org/string-functions#encodeurl
           URL url = new URL(str);
@@ -1083,7 +1104,7 @@ public static List<String> split(String str, Object pattern, Number limit) {
                 // $split("str", ""): Split string into characters
                 int l = limit!=null ? limit.intValue() : Integer.MAX_VALUE;
                 for (int i=0; i<str.length() && i<l; i++) {
-                    result.add( ""+str.charAt(i) );
+                    result.add(String.valueOf(str.charAt(i)));
                 }
             } else {
                 // Quote separator string + preserve trailing empty strings (-1)