Merge pull request #51 from code4craft/test

Update RegexSelector.java

webmagic-core/src/main/java/us/codecraft/webmagic/selector/RegexSelector.java

@@ -26,12 +26,12 @@ public class RegexSelector implements Selector {
         if (StringUtils.isBlank(regexStr)) {
             throw new IllegalArgumentException("regex must not be empty");
         }
-        if (!StringUtils.contains(regexStr, "(") && !StringUtils.contains(regexStr, ")")) {
+        // Check bracket for regex group. Add default group 1 if there is no group.
+        // Only check if there exists the valid left parenthesis, leave regexp validation for Pattern.
+        if (StringUtils.countMatches(regexStr, "(") - StringUtils.countMatches(regexStr, "\\(") ==
+                StringUtils.countMatches(regexStr, "(?:") - StringUtils.countMatches(regexStr, "\\(?:")) {
             regexStr = "(" + regexStr + ")";
         }
-        if (!StringUtils.contains(regexStr, "(") || !StringUtils.contains(regexStr, ")")) {
-            throw new IllegalArgumentException("regex must have capture group 1");
-        }
         this.regexStr = regexStr;
         try {
             regex = Pattern.compile(regexStr, Pattern.DOTALL | Pattern.CASE_INSENSITIVE);

webmagic-core/src/test/java/us/codecraft/webmagic/selector/RegexSelectorTest.java

 package us.codecraft.webmagic.selector;
 
-import junit.framework.Assert;
+import org.junit.Assert;
 import org.junit.Test;
 
 /**
@@ -8,14 +8,18 @@ import org.junit.Test;
  */
 public class RegexSelectorTest {
 
-    @Test
-    public void testInvalidRegex() {
+    @Test(expected = IllegalArgumentException.class)
+    public void testRegexWithSingleLeftBracket() {
         String regex = "\\d+(";
-        try {
         new RegexSelector(regex);
-            Assert.assertNotNull(regex);
-        } catch (Exception e) {
-
     }
+
+    @Test
+    public void testRegexWithLeftBracketQuoted() {
+        String regex = "\\(.+";
+        String source = "(hello world";
+        RegexSelector regexSelector = new RegexSelector(regex);
+        String select = regexSelector.select(source);
+        Assert.assertEquals(source,select);
     }
 }