Improve regex to avoid backtrack and to use non-capturing groups (#110)

1 files changed, 20 insertions, 15 deletions

diff --git a/sonar-css-plugin/src/main/java/org/sonar/css/plugin/CssLexer.java b/sonar-css-plugin/src/main/java/org/sonar/css/plugin/CssLexer.java
index 9a4bb58..7aa1200 100644
--- a/sonar-css-plugin/src/main/java/org/sonar/css/plugin/CssLexer.java
+++ b/sonar-css-plugin/src/main/java/org/sonar/css/plugin/CssLexer.java
@@ -27,33 +27,36 @@ import static com.sonar.sslr.impl.channel.RegexpChannelBuilder.regexp;
 // It is far from being entirely matching the standard definition of css/less/scss tokens nor
 // following the theory of what a lexer responsibilities are but as we are only building line metrics and highlighting
 // on top of it we decided to focus on simplicity over being extensive.
+
+// Be careful to avoid/limit usage of backtracking regex. There is nearly always an alternative with a forward lookup.
+// This will allow to improve performance and avoid a lof of StackOverflowException.
 public final class CssLexer {
 
-  private static final String NEW_LINE = "(\r\n|\r|\n|\f)";
+  private static final String NEW_LINE = "(?:\r\n|\r|\n|\f)";
   private static final String WHITESPACE = "[\t\n\f\r ]";
   private static final String NON_ASCII = "[^\\p{ASCII}]";
   private static final String HEX_DIGIT = "0-9a-fA-F";
-  private static final String ESCAPE = "(\\\\[" + HEX_DIGIT + "]{1,6}" + WHITESPACE + "?)|\\[^\r\n\f" + HEX_DIGIT + "]";
+  private static final String ESCAPE = "(?:\\\\[" + HEX_DIGIT + "]{1,6}" + WHITESPACE + "?)|\\[^\r\n\f" + HEX_DIGIT + "]";
 
   private static final String PUNCTUATOR = "[!:,;%&+#\\*-/=>\\(\\)\\[\\]\\{\\}]";
 
-  private static final String MULTI_LINE_COMMENT = "/\\*(.|" + NEW_LINE + ")*?\\*/";
-  private static final String INLINE_COMMENT = "//.*";
-  private static final String COMMENT  = "(" + INLINE_COMMENT + "|" + MULTI_LINE_COMMENT + ")";
+  // Use dotall mode (https://docs.oracle.com/javase/8/docs/api/java/util/regex/Pattern.html#DOTALL) to match line return
+  // while using .
+  private static final String MULTI_LINE_COMMENT = "(?s)/\\*.*?\\*/";
+  private static final String INLINE_COMMENT = "//[^\n\r\f]*+";
 
-  private static final String NUMBER = "[+|-]?\\d*\\.?\\d+([a-z]+|%)?";
+  private static final String NUMBER = "[+|-]?+(?:\\d++(?:.\\d++)?+|\\.\\d++)(?:[a-z]++|%)?+";
 
   private static final String NAME_CHAR = "[a-zA-Z0-9_-]|" + NON_ASCII + "|" + ESCAPE;
   private static final String NAME_START = "[a-zA-Z_]|" + NON_ASCII + "|" + ESCAPE;
 
-  private static final String IDENTIFIER = "-?(" + NAME_START + ")(" + NAME_CHAR + ")*";
-  private static final String AT_IDENTIFIER = "@+" + IDENTIFIER;
-  private static final String HASH_IDENTIFIER = "#(" + NAME_CHAR + ")+";
-  private static final String DOLLAR_IDENTIFIER = "\\$(" + NAME_CHAR + ")+";
+  private static final String IDENTIFIER = "-?+(?:" + NAME_START + ")(?:" + NAME_CHAR + ")*+";
+  private static final String AT_IDENTIFIER = "@++" + IDENTIFIER;
+  private static final String HASH_IDENTIFIER = "#(?:" + NAME_CHAR + ")++";
+  private static final String DOLLAR_IDENTIFIER = "\\$(?:" + NAME_CHAR + ")++";
 
-  private static final String DOUBLE_QUOTE_STRING = "~?\"([^\"\\\\\r\n\f]|" + ESCAPE + "|\\\\" + NEW_LINE + ")*\"";
-  private static final String SINGLE_QUOTE_STRING = "~?'([^'\\\\\r\n\f]|" + ESCAPE + "|\\\\" + NEW_LINE + ")*'";
-  private static final String STRING = "(" + SINGLE_QUOTE_STRING + "|" + DOUBLE_QUOTE_STRING + ")";
+  private static final String DOUBLE_QUOTE_STRING = "~?+\"(?:[^\"\\\\\r\n\f]|" + ESCAPE + "|\\\\" + NEW_LINE + ")*+\"";
+  private static final String SINGLE_QUOTE_STRING = "~?+'(?:[^'\\\\\r\n\f]|" + ESCAPE + "|\\\\" + NEW_LINE + ")*+'";
 
   private CssLexer() {
   }
@@ -62,8 +65,10 @@ public final class CssLexer {
     return Lexer.builder()
       .withFailIfNoChannelToConsumeOneCharacter(false)
 
-      .withChannel(regexp(CssTokenType.COMMENT, COMMENT))
-      .withChannel(regexp(CssTokenType.STRING, STRING))
+      .withChannel(regexp(CssTokenType.COMMENT, MULTI_LINE_COMMENT))
+      .withChannel(regexp(CssTokenType.COMMENT, INLINE_COMMENT))
+      .withChannel(regexp(CssTokenType.STRING, DOUBLE_QUOTE_STRING))
+      .withChannel(regexp(CssTokenType.STRING, SINGLE_QUOTE_STRING))
       .withChannel(regexp(CssTokenType.AT_IDENTIFIER, AT_IDENTIFIER))
       .withChannel(regexp(CssTokenType.HASH_IDENTIFIER, HASH_IDENTIFIER))
       .withChannel(regexp(CssTokenType.DOLLAR_IDENTIFIER, DOLLAR_IDENTIFIER))