[Java] Improve cucumber expression creation performance (#202)

Co-authored-by: M.P. Korstanje <[email protected]>
Co-authored-by: M.P. Korstanje <[email protected]>

Author: jkronegg

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](http://semver.org/).
 
 ## [Unreleased]
 
+### Fixed
+- [Java] Improve cucumber expression creation performance ([#202](https://github.com/cucumber/cucumber-expressions/pull/202))
+
 ## [16.1.1] - 2022-12-08
 ### Fixed
 - [Java] Improve expression creation performance ([#187](https://github.com/cucumber/cucumber-expressions/pull/187), [#189](https://github.com/cucumber/cucumber-expressions/pull/189))

java/src/main/java/io/cucumber/cucumberexpressions/CucumberExpression.java

@@ -19,12 +19,12 @@
 import static io.cucumber.cucumberexpressions.CucumberExpressionException.createOptionalMayNotBeEmpty;
 import static io.cucumber.cucumberexpressions.CucumberExpressionException.createParameterIsNotAllowedInOptional;
 import static io.cucumber.cucumberexpressions.ParameterType.isValidParameterTypeName;
+import static io.cucumber.cucumberexpressions.RegexpUtils.escapeRegex;
 import static io.cucumber.cucumberexpressions.UndefinedParameterTypeException.createUndefinedParameterType;
 import static java.util.stream.Collectors.joining;
 
 @API(status = API.Status.STABLE)
 public final class CucumberExpression implements Expression {
-    private static final Pattern ESCAPE_PATTERN = Pattern.compile("[\\\\^\\[({$.|?*+})\\]]");
     private final List<ParameterType<?>> parameterTypes = new ArrayList<>();
     private final String source;
     private final TreeRegexp treeRegexp;
@@ -60,11 +60,6 @@ private String rewriteToRegex(Node node) {
         }
     }
 
-    private static String escapeRegex(String text) {
-        return ESCAPE_PATTERN.matcher(text).replaceAll("\\\\$0");
-    }
-
-    
     private String rewriteOptional(Node node) {
         assertNoParameters(node, astNode -> createParameterIsNotAllowedInOptional(astNode, source));
         assertNoOptionals(node, astNode -> createOptionalIsNotAllowedInOptional(astNode, source));

java/src/main/java/io/cucumber/cucumberexpressions/RegexpUtils.java

@@ -0,0 +1,72 @@
+package io.cucumber.cucumberexpressions;
+
+class RegexpUtils {
+    /**
+     * List of characters to be escaped.
+     * The last char is '}' with index 125, so we need only 126 characters.
+     */
+    private static final boolean[] CHAR_TO_ESCAPE = new boolean[126];
+
+    static {
+        CHAR_TO_ESCAPE['^'] = true;
+        CHAR_TO_ESCAPE['$'] = true;
+        CHAR_TO_ESCAPE['['] = true;
+        CHAR_TO_ESCAPE[']'] = true;
+        CHAR_TO_ESCAPE['('] = true;
+        CHAR_TO_ESCAPE[')'] = true;
+        CHAR_TO_ESCAPE['{'] = true;
+        CHAR_TO_ESCAPE['}'] = true;
+        CHAR_TO_ESCAPE['.'] = true;
+        CHAR_TO_ESCAPE['|'] = true;
+        CHAR_TO_ESCAPE['?'] = true;
+        CHAR_TO_ESCAPE['*'] = true;
+        CHAR_TO_ESCAPE['+'] = true;
+        CHAR_TO_ESCAPE['\\'] = true;
+    }
+
+    /**
+     * Escapes the regexp characters (the ones from "^$(){}[].+*?\")
+     * from the given text, so that they are not considered as regexp
+     * characters.
+     *
+     * @param text the non-null input text
+     * @return the input text with escaped regexp characters
+     */
+    public static String escapeRegex(String text) {
+        /*
+        Note on performance: this code has been benchmarked for
+        escaping frequencies of 100%, 50%, 20%, 10%, 1%, 0.1%.
+        Amongst 4 other variants (including Pattern matching),
+        this variant is the faster on all escaping frequencies.
+        */
+        int length = text.length();
+        StringBuilder sb = null; // lazy initialization
+        int blockStart = 0;
+        int maxChar = CHAR_TO_ESCAPE.length;
+        for (int i = 0; i < length; i++) {
+            char currentChar = text.charAt(i);
+            if (currentChar < maxChar && CHAR_TO_ESCAPE[currentChar]) {
+                if (sb == null) {
+                    sb = new StringBuilder(length * 2);
+                }
+                if (blockStart < i) {
+                    // flush previous block
+                    sb.append(text, blockStart, i);
+                }
+                sb.append('\\');
+                sb.append(currentChar);
+                blockStart = i + 1;
+            }
+        }
+        if (sb != null) {
+            // finalizing character escaping
+            if (blockStart < length) {
+                // flush remaining characters
+                sb.append(text, blockStart, length);
+            }
+            return sb.toString();
+        }
+        return text;
+    }
+
+}

java/src/test/java/io/cucumber/cucumberexpressions/RegexpUtilsTest.java

@@ -0,0 +1,36 @@
+package io.cucumber.cucumberexpressions;
+
+import org.junit.jupiter.api.Test;
+
+import static io.cucumber.cucumberexpressions.RegexpUtils.escapeRegex;
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+class RegexpUtilsTest {
+
+    @Test
+    void escape_regex_characters(){
+        assertEquals("hello \\$world", escapeRegex("hello $world"));
+    }
+
+    @Test
+    void escape_all_regexp_characters() {
+        assertEquals("\\^\\$\\[\\]\\(\\)\\{\\}\\.\\|\\?\\*\\+\\\\", escapeRegex("^$[](){}.|?*+\\"));
+    }
+
+    @Test
+    void escape_escaped_regexp_characters() {
+        assertEquals("\\^\\$\\[\\]\\\\\\(\\\\\\)\\{\\}\\\\\\\\\\.\\|\\?\\*\\+", escapeRegex("^$[]\\(\\){}\\\\.|?*+"));
+    }
+
+
+    @Test
+    void do_not_escape_when_there_is_nothing_to_escape() {
+        assertEquals("hello world", escapeRegex("hello world"));
+    }
+
+    @Test
+    void gives_no_error_for_unicode_characters() {
+        assertEquals("🥒", escapeRegex("🥒"));
+    }
+
+}